team-repair
Skill145 starsChanged 22 days agoMITView on GitHub
Use when indexes are inconsistent, orphans are detected, after bulk changes to team-ai-directives, or for periodic directives health validation; --build-to-delete proposes rules the model no longer needs; --validate-drafts validates draft files in .adlc/drafts/ without modifying them.
In this repository48 other files it gives its agents.
- architect-analyzeSkill
- architect-bootSkill
- architect-clarifySkill
- architect-implementSkill
- architect-initSkill
- architect-specifySkill
- writing-skillsSkill
- change-bootSkill
- change-clarifySkill
- change-initSkill
- change-publishSkill
- change-specifySkill
- evals-analyzeSkill
- evals-clarifySkill
- evals-implementSkill
- evals-initSkill
- evals-specifySkill
- evals-validateSkill
- factory-cleanSkill
- factory-initSkill
- factory-intentSkill
- factory-learnSkill
- factory-missionSkill
- factory-orchestratorSkill
- factory-queueSkill
- factory-releaseSkill
- factory-reviewSkill
- factory-runbookSkill
- factory-setupSkill
- factory-ticketsSkill
- mission-briefSkill
- product-analyzeSkill
- product-bootSkill
- product-clarifySkill
- product-implementSkill
- product-initSkill
- product-roadmapSkill
- product-specifySkill
- team-bootSkill
- team-constitutionSkill
- team-diagnoseSkill
- team-discoverSkill
- team-initSkill
- team-levelupSkill
- team-setupSkill
- team-skillsSkill
- workspaceSkill
- tech-radar-bootSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
59 rules. About 0 are common, 23 are distinctive, and 36 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈23
Distinctive · estimate
This author's own choices. Worth a look.
36
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 41
Reporting: all changes made
- Very specific: line 95
Objective: Run a non-destructive health check against the team directives framework before proceeding with repairs.
- Very specific: line 102
Verify team_ai_directives field exists and points to valid path
- Distinctive: line 103
Check the team AI directives path exists
- Very specific: line 109
Read .adlc/init-options.json → get team AI directives path
- Very specific: line 135
Read team constitution from {TEAM_AI_DIRECTIVES}/context_modules/constitution.md
- Distinctive: line 137
If project constitution exists:
- Distinctive: line 138
Check if it references team-ai-directives (e.g., "Based on team-ai-directives", "Inherits from")
- Distinctive: line 139
Check if team principles are present in project constitution (compare principle titles)
- Distinctive: line 143
If project constitution doesn't exist:
- Very specific: line 150
Verify type field is present and has a valid value:
- Distinctive: line 152
Verify OKF v0.2 fields (migrate if v0.1 detected):
- Very specific: line 163
Read {REPO_ROOT}/AGENTS.md (the project-level agent instructions file)
- Very specific: line 164
Check if it contains the <!-- TEAM_AI_DIRECTIVES START --> marker
- Distinctive: line 165
If the marker exists, verify the managed section includes:
- Very specific: line 189
If missing, create it (call setup-team.sh --ensure-adlc or create inline)
- Very specific: line 204
If any check is [FAIL], print the report, set exit code 1, and STOP.
- Very specific: line 220
Run bash scripts/bash/setup-team.sh --json (or pwsh scripts/powershell/setup-team.ps1 -Json) to resolve paths and parse JSON output:
- Distinctive: line 234
Objective: Ensure team-ai-directives is configured
- Very specific: line 236
Check if TEAM_AI_DIRECTIVES has a value from script output.
- Distinctive: line 247
Objective: Ensure AGENTS.md exists with required structure
- Distinctive: line 267
Check for each required section:
- Very specific: line 311
After repairing the team AI directives' own AGENTS.md, also ensure the project-level AGENTS.md (at {REPO_ROOT}/AGENTS.md) contains the …
- Very specific: line 313
If Check 7 returned [WARN] or [INFO], run the injection:
- Very specific: line 382
Check if starts with ---
- Distinctive: line 388
Before generating new frontmatter, read the existing CDR.md to find pre-existing CDR references for orphan files.
- Very specific: line 418
Remove .md extension, replace / with -, prepend type prefix
- Very specific: line 426
If found, use the existing cdr_ref
- Very specific: line 427
If not found, set cdr_ref: null
- Distinctive: line 461
Write back to file
- Very specific: line 542
Check SKILL.md exists (required)
- Very specific: line 543
Check .skills-entry.json exists (optional)
- Very specific: line 550
Categories: Look for ## Categories or ## Trigger Keywords section
- Very specific: line 551
Instruction Type: Look for Instruction Type: line
- Very specific: line 788
Run the conflict scan over {TEAM_AI_DIRECTIVES}/context_modules/rules inline (no cross-skill helper — scripts stay skill-local), assigning …
- Very specific: line 834
Objective: Update verified timestamps for valid directives and flag stale ones.
- Distinctive: line 841
If no conflicts detected for this file → eligible for verification update
- Distinctive: line 842
If conflicts detected → skip (will be resolved via conflict CDRs)
- Very specific: line 850
Update generated.at if content changed during this repair run
- Very specific: line 873
Objective: Read usage data from adlc orphan branch, aggregate into confidence scores, and update OKF frontmatter on main branch.
- Very specific: line 895
Write to adlc branch: reports/confidence-scores.json.
- Very specific: line 901
Add/update confidence: block in YAML frontmatter
- Distinctive: line 902
Commit to main branch
- Very specific: line 906
Add Confidence and Usage columns to the derived CDR.md table.
- Very specific: line 918
Read all goldenset directories from {TEAM_AI_DIRECTIVES}/evals/:
- Distinctive: line 928
If no goldensets exist, report: "No evals found — run /team-levelup to create eval CDRs first." and skip this phase.
- Very specific: line 933
Read paired_directive from the goldenset frontmatter
- Very specific: line 934
Read the directive file from context_modules/ (e.g., rules/security/sql_injection_prevention.md)
- Distinctive: line 935
If the directive file doesn't exist → skip (already deleted or orphaned eval)
- Very specific: line 1081
If no draft files are found in any directory, report: "No draft files found in .adlc/drafts/ — nothing to validate." and skip this phase.
- Distinctive: line 1103
After frontmatter, validate that the following required body sections are present as Markdown headings:
- Distinctive: line 1122
No content between the heading and the next heading or end of file
- Distinctive: line 1123
Only whitespace or placeholder text (e.g., "N/A", "TBD", "TODO", "none")
- Very specific: line 1164
If errors were found: suggest fixing the draft files before promoting them via the appropriate clarify skill (/architect-clarify for ADRs, …
- Distinctive: line 1165
If all drafts pass validation: confirm drafts are structurally ready for promotion.
- Distinctive: line 1254
Auto-fix: Always repairs issues automatically (no confirmation needed)
- Very specific: line 1255
Dry run: Use --dry-run to preview changes without writing
- Very specific: line 1256
Selective repair: Use --index-only, --skills-only, or --agents-only for specific targets
- Very specific: line 1310
Default fallback: team-ai-directives/ relative to project root.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
team-repair
Scope rule: scripts stay skill-local — never source a script from another skill. For another skill's operations, invoke that skill (e.g.
team-setup,team-levelup,architect-clarify) via the Skill tool; for its documents, follow theRequiresline.scripts/andreferences/below are relative to this skill.{REPO_ROOT}denotes the project root (where.adlc/lives).
Requires: the team-setup skill for AGENTS.md injection and team configuration: adlc-cli skills add tikalk/adlc-team-skills --skill team-setup
Overview
Re-indexes OKF v0.2 artifacts (index.md, log.md), derives the flat CDR.md inject, rebuilds .skills.json and AGENTS.md in team-ai-directives to fix inconsistencies, detect orphaned files, and auto-repair issues. Migrates OKF v0.1 frontmatter to v0.2 on every run (always-on, no opt-out). Begins with a health-check phase (Phase 0) that verifies the directives framework is installed, configured, and aligned before performing any repairs.
Input: team-ai-directives repository
Output: 0. Health check report (7 checks: team AI directives configured, context modules exist, skills registry, OKF log tracking, constitution alignment, OKF v0.2 type field presence, project AGENTS.md directive)
- Repaired AGENTS.md (if missing or corrupted)
- Migrated all context module frontmatter from OKF v0.1 to v0.2 (always-on)
- Rebuilt per-directory
index.mdfiles (OKF §8 catalog) - Rebuilt per-directory
log.mdfiles (OKF §9 audit trail) - Derived
CDR.mdflat table fromindex.mdfiles (for team-boot system prompt injection) - Rebuilt .skills.json manifest from skills/
- Auto-added OKF v0.2 YAML frontmatter to orphan context modules
- Auto-generated .skills.json entries for orphan skills
- Conflict scan across rules (creates conflict CDRs if issues found)
- Freshness verification (updates
verifiedtimestamps, flags stale directives) - Draft validation report (if
--validate-drafts— read-only, no modifications) - Summary report of all repairs
You are acting as an Index Repair Specialist ensuring team-ai-directives indexes are consistent and complete. Your role involves:
- Verifying health checks before repair (Phase 0)
- Scanning context_modules/ and skills/ directories
- Detecting orphan files (missing frontmatter/manifest entries)
- Auto-repairing issues by generating missing metadata
- Rebuilding index files to reflect actual content
- Reporting all changes made
Repair Targets
| Target | Location | Purpose |
| AGENTS.md | {TEAM_AI_DIRECTIVES}/AGENTS.md | Main instruction file for AI agents |
| index.md | {TEAM_AI_DIRECTIVES}/context_modules/**/index.md | OKF §8 per-directory catalogs (progressive disclosure) |
| log.md | {TEAM_AI_DIRECTIVES}/context_modules/**/log.md | OKF §9 per-directory audit trails |
| CDR.md | {TEAM_AI_DIRECTIVES}/CDR.md | Derived flat table for team-boot system prompt injection (auto-generated from index.md files) |
| .skills.json | {TEAM_AI_DIRECTIVES}/.skills.json | Skills manifest registry |
When to Use
User Input
$ARGUMENTSYou MUST consider the user input before proceeding (if not empty).
Examples of User Input:
""- Repair all three indexes (default)"--dry-run"- Report only, don't write changes"--index-only"- Only rebuild OKF index.md + log.md + derive CDR.md"--skills-only"- Only repair .skills.json"--agents-only"- Only repair AGENTS.md"--validate-drafts"- Validate draft files in .adlc/drafts/ without modifying them- Empty input: Repair all indexes with auto-fix (includes v0.1→v0.2 migration)
Flags
| Flag | Description |
--dry-run | Report only, don't write changes |
--health-only | Run Phase 0 health check only, then stop. |
--validate | Run conflict scan + freshness verification only (Phases 8-9) |
--conflicts | Scan for rule conflicts only |
--freshness | Verify directive freshness only |
--build-to-delete | Run evals without directives to identify candidates for removal (Factor XII) |
--validate-drafts | Validate draft files in .adlc/drafts/{adr,pdr,chdr,cdr,evals}/ — validation only, no modifications |
--update-confidence | Aggregate usage data from adlc branch and update OKF frontmatter confidence scores |
--ensure-adlc | Ensure the adlc orphan branch exists (create if missing) |
--index-only | Only rebuild OKF index.md + log.md + derive CDR.md |
--skills-only | Only repair .skills.json |
--agents-only | Only repair AGENTS.md |
| (default) | Repair all indexes + migrate v0.1→v0.2 + validate conflicts and freshness |
Core Process
Phase 0: Health Check
Objective: Run a non-destructive health check against the team directives framework before proceeding with repairs. If any check returns [FAIL], present the report and stop — the framework is not healthy enough to repair safely.
Execute all eight checks below. Each check prints a status line. If any check is [FAIL], abort repair.
Check 1: Team AI Directives Configured
- Read
.adlc/init-options.json - Verify
team_ai_directivesfield exists and points to valid path - Check the team AI directives path exists
Output: [OK] or [FAIL] with reason
Check 2: Context Modules Exist
- Read
.adlc/init-options.json→ get team AI directives path - Verify:
{TEAM_AI_DIRECTIVES}/context_modules/constitution.md{TEAM_AI_DIRECTIVES}/context_modules/personas/{TEAM_AI_DIRECTIVES}/context_modules/rules/{TEAM_AI_DIRECTIVES}/context_modules/examples/
Output: [OK] or [FAIL] with reason
Check 3: Skills Registry
{TEAM_AI_DIRECTIVES}/.skills.jsonexists and is valid JSON
Output: [OK] or [FAIL] with reason
Check 4: OKF Log Tracking
{TEAM_AI_DIRECTIVES}/context_modules/rules/log.mdexists{TEAM_AI_DIRECTIVES}/context_modules/personas/log.mdexists{TEAM_AI_DIRECTIVES}/context_modules/examples/log.mdexists{TEAM_AI_DIRECTIVES}/CDR.mdexists (derived artifact)
Output: [OK] or [FAIL] with reason
Check 5: Constitution Alignment
- Read team constitution from
{TEAM_AI_DIRECTIVES}/context_modules/constitution.md - Locate project constitution: the project root (where
.adlc/lives) →{REPO_ROOT}/docs/adlc/memory/constitution.md, falling back to legacy{REPO_ROOT}/.adlc/memory/constitution.md(dual-read) - If project constitution exists:
- Check if it references team-ai-directives (e.g., "Based on team-ai-directives", "Inherits from")
- Check if team principles are present in project constitution (compare principle titles)
- Output:
[OK]— Project constitution exists and inherits team principles[WARN]— Project constitution exists but missing team inheritance
- If project constitution doesn't exist:
[INFO]— Project constitution doesn't exist yet (first-time setup)
Check 6: OKF v0.2 Conformance
- Scan all
.mdfiles incontext_modules/(excludingindex.md,log.md) - Parse YAML frontmatter from each file
- Verify
typefield is present and has a valid value:- Valid types:
Constitution,Persona,Rule,Example,Skill
- Valid types:
- Verify OKF v0.2 fields (migrate if v0.1 detected):
generated: { by, at }present (not legacytimestamp)verifiedis a list format (not bare string)statuspresent (e.g.,stable,draft,deprecated)stale_afterpresent (e.g.,180d)
- Output:
[OK]— All concept files have validtypefields and v0.2 families[WARN]— Some files missing v0.2 fields or still carry v0.1 fields (will be migrated in Phase 4)
Check 7: Project AGENTS.md Directive
- Read
{REPO_ROOT}/AGENTS.md(the project-level agent instructions file) - Check if it contains the
<!-- TEAM_AI_DIRECTIVES START -->marker - If the marker exists, verify the managed section includes:
- A
team-bootinvocation directive - A reference to team AI directives context (constitution, CDR index)
- The Class Boots catalog (architect-boot / product-boot / change-boot / team-boot / tech-radar-boot)
- The compact Decision Capture triggers + Team Context & Decisions contract (single merged table; Status carries the accepted-vs-pending distinction)
- A
- Output:
[OK]— Project AGENTS.md contains a valid team AI directives managed section[WARN]— Project AGENTS.md exists but is missing the managed section (agents won't auto-invoketeam-boot)[INFO]— Project AGENTS.md doesn't exist yet (first-time setup)
Check 8: Deterministic Enforcement Coverage
Advisory check — outputs [OK]/[WARN], never [FAIL] (deterministic-checks-first, EVAL-010). A missing guardrail is a finding on its own, not just a mistake's side effect.
- Scan rule CDRs in
{TEAM_AI_DIRECTIVES}/context_modules/rules/for mechanically-checkable patterns — fixed syntactic shapes, banned APIs, import shapes, file-location rules — that lack a paired deterministic check (unit test, binary grader, pre-commit hook, lint rule, or CI job) - Scan installed skills for missing eval coverage — a skill with neither a goldset criterion/grader nor a stated no-grader reason has no guardrail
Output:
[OK]— every mechanical rule has a paired check; every skill has eval coverage or a stated reason[WARN]— N mechanical rules lack checks; M skills lack eval coverage (promotion candidates → feed to factory-learn Maintenance route / team-levelup Phase 2b, action P)
Check 9: adlc Orphan Branch
- Check if
git -C "$TEAM_AI_DIRECTIVES" show-ref --verify --quiet refs/heads/adlc - If missing, create it (call
setup-team.sh --ensure-adlcor create inline)
Output:
[OK]— adlc orphan branch exists withdrafts/cdr/andreports/structure[WARN]— adlc branch missing (auto-created during repair)[INFO]— adlc branch created
Health Check Output
Print verification status for each check:
[OK]— Check passed[FAIL]— Check failed with reason (abort repair)[WARN]— Check passed with warnings (non-blocking)[INFO]— Informational only
If any check is [FAIL], print the report, set exit code 1, and STOP. Do not proceed to Phase 1.
Health Check Red Flags
[FAIL]on Check 1 or Check 2: the directives framework is effectively absent — agents have nothing to inherit from. Stop and reinstall before repairing.- Team AI Directives path resolves outside the repo or to a temp/scratch location: the project is pointing at a transient or shared team AI directives that may vanish or diverge.
{TEAM_AI_DIRECTIVES}/.skills.jsonis missing or not valid JSON: skill discovery is broken; agents cannot find team skills even if the files exist.- Project constitution exists but shows no team inheritance (
[WARN]on Check 5): the project was bootstrapped without the team AI directives, or the constitution was hand-edited and the inheritance markers were removed. - Multiple checks return
[WARN]simultaneously: systemic drift, usually from a moved.adlc/directory or a reconfigured team AI directives path. Treat as a[FAIL]-equivalent and re-init.
Phase 1: Environment Setup
Objective: Resolve paths and validate infrastructure
Run bash scripts/bash/setup-team.sh --json (or pwsh scripts/powershell/setup-team.ps1 -Json) to resolve paths and parse JSON output:
{
"REPO_ROOT": "/path/to/project",
"TEAM_AI_DIRECTIVES": "/path/to/team-ai-directives",
"BRANCH": "current-branch"
}{REPO_ROOT} is the project root (where .adlc/ lives). Subsequent references use {REPO_ROOT}.
Phase 2: Validate Environment
Objective: Ensure team-ai-directives is configured
Check if TEAM_AI_DIRECTIVES has a value from script output.
If empty, STOP:
Team AI directives repository not configured.
Run: /team-setup
Or set: export TEAM_AI_DIRECTIVES=/path/to/team-ai-directivesPhase 3: Repair AGENTS.md
Objective: Ensure AGENTS.md exists with required structure
Skip if: --index-only or --skills-only flag provided
Step 1: Check AGENTS.md Exists
test -f "{TEAM_AI_DIRECTIVES}/AGENTS.md" && echo "EXISTS" || echo "MISSING"Step 2: Validate Structure (if exists)
Required sections:
# Agent Instructions(title)## Structure## Loading Order## Functional Categories (Rules)## Using Skills## CDR.md
Check for each required section:
grep -q "^# Agent Instructions" "{TEAM_AI_DIRECTIVES}/AGENTS.md"
grep -q "^## Structure" "{TEAM_AI_DIRECTIVES}/AGENTS.md"
grep -q "^## Loading Order" "{TEAM_AI_DIRECTIVES}/AGENTS.md"
grep -q "^## Functional Categories" "{TEAM_AI_DIRECTIVES}/AGENTS.md"
grep -q "^## Using Skills" "{TEAM_AI_DIRECTIVES}/AGENTS.md"
grep -qiE "##.*CDR\.md" "{TEAM_AI_DIRECTIVES}/AGENTS.md"Step 3: Auto-Repair
| Status | Action |
| Missing | Create from ../templates/agents-template.md |
| Corrupted (missing sections) | Overwrite with template |
| Valid | No changes |
If --dry-run:
### AGENTS.md Status: {MISSING|CORRUPTED|VALID}
**Action**: {Would create|Would overwrite|No changes needed}Otherwise, execute repair:
cp "../templates/agents-template.md" "{TEAM_AI_DIRECTIVES}/AGENTS.md"Step 4: Track Results
Store for summary:
{
"agents_md": {
"status": "VALID|CREATED|OVERWRITTEN",
"action": "No changes|Created from template|Re-created from template"
}
}Step 5: Inject Project-Level AGENTS.md Directive
After repairing the team AI directives' own AGENTS.md, also ensure the project-level AGENTS.md (at {REPO_ROOT}/AGENTS.md) contains the team-boot strict-compliance directive. This is what tells agents to invoke team-boot at session start.
If Check 7 returned [WARN] or [INFO], run the injection:
Invoke the team-setup skill (Requires above) to inject the managed section
into {REPO_ROOT}/AGENTS.md (its --inject-agents operation). scripts stay
skill-local, so team-repair never runs team-setup's script directly.
If --dry-run:
### Project AGENTS.md Status: {WARN|INFO}
**Action**: Would inject team AI directives managed section into {REPO_ROOT}/AGENTS.mdOtherwise, execute the injection. The function is idempotent — if the managed section already exists (between <!-- TEAM_AI_DIRECTIVES START --> and <!-- TEAM_AI_DIRECTIVES END --> markers), it replaces the section in place rather than duplicating.
Store for summary:
{
"project_agents_md": {
"status": "VALID|INJECTED|UPDATED",
"action": "No changes|Created with managed section|Updated managed section"
}
}Phase 4: Scan Context Modules for CDR.md Reindex
Objective: Find all context modules and extract metadata
Skip if: --skills-only or --agents-only flag provided
Step 1: Find All Context Module Files
find "{TEAM_AI_DIRECTIVES}/context_modules/rules" -name "*.md" -type f 2>/dev/null
find "{TEAM_AI_DIRECTIVES}/context_modules/personas" -name "*.md" -type f 2>/dev/null
find "{TEAM_AI_DIRECTIVES}/context_modules/examples" -name "*.md" -type f 2>/dev/nullSkip constitution.md (not indexed in CDR.md).
Step 2: Extract YAML Frontmatter
For each file, parse YAML frontmatter (OKF v0.2 form after migration):
---
type: Rule
title: Python error handling
description: Python error handling patterns and best practices
tags: [python, error-handling]
resource: ./context_modules/rules/python/error-handling.md
generated: { by: agent:legacy, at: 2026-04-15T00:00:00Z }
id: rule-python-error-handling
cdr_ref: CDR-2026-001
created: 2026-04-15
verified:
- { by: process:team-repair, at: 2026-05-18T00:00:00Z }
status: stable
stale_after: 180d
sources:
- id: commit-abc123
resource: src/errors.py
title: Error handling implementation
---Extraction logic:
- Read file content
- Check if starts with
--- - Parse YAML between
---markers - Extract:
id,cdr_ref,created,type,title,description,tags,generated.at,verified(latestat),status
Step 2a: Build CDR Lookup from Existing CDR.md
Before generating new frontmatter, read the existing CDR.md to find pre-existing CDR references for orphan files.
Parse the CDR.md index table to build a mapping of {relative_file_path → cdr_ref}:
# Read existing CDR.md and extract file path -> CDR reference mappings
CDR_LOOKUP=()
if [[ -f "{TEAM_AI_DIRECTIVES}/CDR.md" ]]; then
while IFS='|' read -r _ id module _ _ _ _ _; do
id="${id// /}"
module="${module// /}"
if [[ -n "$id" && -n "$module" && "$id" =~ ^CDR- ]]; then
CDR_LOOKUP["$module"]="$id"
fi
done < <(grep "| CDR-" "{TEAM_AI_DIRECTIVES}/CDR.md")
fiThis creates an associative array:
context_modules/rules/style-guides/java/google_style_guide.md → CDR-2026-023Step 3: Detect Orphans (No Frontmatter)
Files with .md extension but no YAML frontmatter.
For each orphan:
- Generate
idfrom filename:- Strip the context type directory prefix (
rules/,personas/,examples/) - Remove
.mdextension, replace/with-, prepend type prefix - Example:
rules/python/new-pattern.md→ striprules/→python/new-pattern.md→rule-python-new-pattern - Example:
personas/architect.md→ strippersonas/→architect.md→persona-architect
- Strip the context type directory prefix (
- Determine context type from path:
rules/→Rulepersonas/→Personaexamples/→Example
- Compute the file's relative path from
TEAM_AI_DIRECTIVESand look it up inCDR_LOOKUP:- If found, use the existing
cdr_ref - If not found, set
cdr_ref: null
- If found, use the existing
- Generate
titlefrom filename (humanize the basename) - Generate
descriptionfrom first paragraph or filename - Generate
tagsfrom path segments (e.g.,rules/python/→[python]) - Set default metadata:
type: {context-type} title: {generated-title} description: {generated-description} tags: {generated-tags} resource: {relative-path} generated: { by: agent:team-repair, at: {today}T00:00:00Z } id: {generated-id} cdr_ref: {from CDR_LOOKUP or null} created: {today} verified: - { by: agent:team-repair, at: {today}T00:00:00Z } status: stable stale_after: 180d
If --dry-run:
### Orphan Files Detected
| File | Generated ID | Existing CDR Ref | Action |
|------|--------------|-----------------|--------|
| rules/python/new-pattern.md | rule-python-new-pattern | CDR-2026-023 | Would add frontmatter (preserving CDR ref) |
| personas/architect.md | persona-architect | null | Would add frontmatter |Otherwise, auto-fix:
- Read file content
- Prepend generated YAML frontmatter
- Write back to file
Step 3b: Migrate v0.1 Frontmatter to v0.2 (Always-On)
For every .md file in context_modules/ (excluding index.md, log.md) that has existing frontmatter, detect and migrate v0.1 fields to v0.2. This runs on every team-repair invocation — no flag, no opt-out.
Migration rules (idempotent — skip if already v0.2):
| Detection | Action |
timestamp: present, generated: absent | Rewrite to generated: { by: agent:legacy, at: <timestamp value> }, delete timestamp |
timestamp: present, generated: present | Delete timestamp (v0.2 takes precedence) |
verified: is a bare string (not a list) | Rewrite to verified: [{ by: process:team-repair, at: <value>T00:00:00Z }] |
evidence: present with entries | Map each entry to sources[] entry (id, resource, title), delete evidence |
evidence: [] (empty list) | Delete field, omit sources |
modified: present | Delete (redundant with generated.at) |
age_days: present | Delete (derived at render time) |
status: absent | Add status: stable |
stale_after: absent | Add stale_after: 180d |
resource: absent | Add from file relative path |
generated already present, no v0.1 fields | Skip (already v0.2) |
Preserve custom fields (id, cdr_ref, created, type, title, description, tags) as-is per OKF §4.1.
If --dry-run:
### v0.1 → v0.2 Migration Preview
| File | Fields Migrated | Fields Added | Fields Removed |
|------|----------------|--------------|----------------|
| rules/security/sql_injection_prevention.md | timestamp→generated, verified→list | status, stale_after, resource | modified, age_days, evidence |Otherwise, rewrite frontmatter in place for each file.
Step 4: Build Context Module Index
Create index structure:
{
"context_modules": [
{
"file": "context_modules/rules/python/error-handling.md",
"id": "rule-python-error-handling",
"cdr_ref": "CDR-2026-001",
"type": "Rule",
"created": "2026-04-15",
"generated_at": "2026-04-15T00:00:00Z",
"verified_at": "2026-05-18T00:00:00Z",
"status": "stable",
"stale_after": "180d",
"descriptor": "Python error handling patterns and best practices"
}
],
"orphans": [
{
"file": "context_modules/rules/python/new-pattern.md",
"id": "rule-python-new-pattern",
"repaired": true
}
]
}Phase 5: Scan Skills for .skills.json Reindex
Objective: Find all skills and build manifest entries
Skip if: --index-only or --agents-only flag provided
Step 1: Find All Skill Directories
find "{TEAM_AI_DIRECTIVES}/skills" -mindepth 1 -maxdepth 1 -type dStep 2: Check Each Skill
For each skill directory:
- Check
SKILL.mdexists (required) - Check
.skills-entry.jsonexists (optional) - Parse SKILL.md for metadata
Step 3: Extract Skill Metadata
From SKILL.md:
- Description: First paragraph after title
- Categories: Look for
## Categoriesor## Trigger Keywordssection - Instruction Type: Look for
**Instruction Type**:line
Step 4: Generate .skills.json Entry
{
"local:./skills/{skill-name}": {
"version": "1.0.0",
"description": "{extracted from SKILL.md first paragraph}",
"categories": ["{from SKILL.md}"],
"instruction_type": "{from SKILL.md}"
}
}Step 5: Detect Orphans
Skills with SKILL.md but no entry in .skills.json.
If --dry-run:
### Orphan Skills Detected
| Skill | Action |
|-------|--------|
| code-review | Would add to .skills.json |
| deployment | Would add to .skills.json |Otherwise, auto-generate entry.
Step 6: Detect Missing Files
Entries in .skills.json where skill directory doesn't exist.
Auto-remove invalid entries.
Step 7: Build Skills Index
{
"skills": [
{
"name": "code-review",
"path": "skills/code-review/",
"has_skill_md": true,
"has_entry": false,
"repaired": true
}
],
"missing_removed": 1
}Phase 6: Rebuild OKF index.md + log.md + Derive CDR.md
Objective: Generate OKF v0.2 per-directory index.md (§8) and log.md (§9) files from scanned context modules, then derive the flat CDR.md table from the index.md data for team-boot system prompt injection.
Skip if: --skills-only or --agents-only flag provided
Step 1: Rebuild Per-Directory index.md (OKF §8)
For each subdirectory (rules/, personas/, examples/) and the context_modules/ root, generate an index.md file in OKF §8 list format.
context_modules/index.md (root — carries okf_version):
---
okf_version: "0.2"
---
# Context Modules
* [Rules](rules/index.md) - Team rules and workflows
* [Personas](personas/index.md) - Team personas
* [Examples](examples/index.md) - Team examplescontext_modules/rules/index.md (per-type catalog):
# Rules
* [Prevent SQL Injection](security/sql_injection_prevention.md) - Standards for preventing SQL injection vulnerabilities across all languages
* [Dependency Injection](architecture/dependency_injection.md) - Dependency injection patterns for maintainable codeEntries are derived from each module's frontmatter title and description. Sort alphabetically by title within each directory. If a module lacks description, derive from first body paragraph.
Personas and Examples follow the same pattern with # Personas and # Examples headings.
Step 2: Rebuild Per-Directory log.md (OKF §9)
For each subdirectory, generate a log.md file in OKF §9 date-grouped format (newest first).
context_modules/rules/log.md:
# Rules Update Log
## 2026-05-23
* **Creation**: Added [Dependency Injection](architecture/dependency_injection.md) — Dependency injection patterns for maintainable code. CDR: CDR-2026-008.
* **Verification**: Verified [SQL Injection Prevention](security/sql_injection_prevention.md). CDR: CDR-2026-021.
## 2026-05-21
* **Creation**: Added [SQL Injection Prevention](security/sql_injection_prevention.md) — Standards for preventing SQL injection. CDR: CDR-2026-021.Log entries are derived from:
- Git history:
git log --diff-filter=A --format="%ai %s" -- <file>for creation dates - Frontmatter
verifiedtimestamps for verification entries - Existing
log.mdcontent (preserve manual entries, append new)
context_modules/log.md (root — aggregate):
# Context Modules Update Log
## 2026-08-12
* **Re-index**: Rebuilt all index.md and log.md files via /team-repair. Rules: {N} files, Personas: {N} files, Examples: {N} files.Step 3: Derive CDR.md (Flat Table for team-boot)
From the per-directory index.md data + module frontmatter, derive a flat CDR.md table for team-boot system prompt injection.
{TEAM_AI_DIRECTIVES}/CDR.md:
# Context Directive Records (Derived Index)
> ⚠️ Auto-generated by `/team-repair`. Do not edit manually.
> Source of truth: `context_modules/*/index.md` + module frontmatter.
> Decision lifecycle (Accepted/Rejected) lives in project `adlc branch drafts/cdr/`.
## CDR Index
| ID | Path | Type | Description | Generated | Verified | Age | Status |
|----|------|------|-------------|-----------|----------|-----|--------|
| CDR-2026-021 | context_modules/rules/security/sql_injection_prevention.md | Rule | Standards for preventing SQL injection... | 2026-06-14 | 2026-05-21 | 46d | stable |
| rule-frontend-routing | context_modules/rules/frontend/framework/frontend_routing.md | Rule | Client-side routing patterns... | 2026-06-15 | 2026-06-15 | 0d | stable |
**Stats**: {N} entries | Last Updated: {date}Columns:
ID:cdr_reffrom frontmatter (oridifcdr_refis null)Path: relative path fromTEAM_AI_DIRECTIVESType:typefrom frontmatterDescription:descriptionfrom frontmatter (truncated to 80 chars)Generated:generated.atdate (YYYY-MM-DD)Verified: latestverified[].atdate (YYYY-MM-DD)Age: days sinceverifieddateStatus:statusfrom frontmatter (orstableif absent)
Step 4: Write Files
If --dry-run:
### OKF Files Preview
Would write:
- context_modules/index.md ({N} entries)
- context_modules/log.md
- context_modules/rules/index.md ({N} entries)
- context_modules/rules/log.md
- context_modules/personas/index.md ({N} entries)
- context_modules/personas/log.md
- context_modules/examples/index.md ({N} entries)
- context_modules/examples/log.md
- CDR.md ({N} derived entries)Otherwise, write all 9 files.
Objective: Generate fresh .skills.json from scanned skills
Skip if: --index-only or --agents-only flag provided
Step 1: Generate Skills Manifest
{
"skills": {
"local:./skills/code-review": {
"version": "1.0.0",
"description": "Review code following team standards and best practices",
"categories": ["review", "quality"],
"instruction_type": "Review"
}
}
}Step 2: Write .skills.json
If --dry-run:
### .skills.json Preview
Would write {N} skill entriesOtherwise:
cat > "{TEAM_AI_DIRECTIVES}/.skills.json" << 'EOF'
{generated JSON}
EOFPhase 8: Conflict Scanning
Objective: Scan team-ai-directives rules for contradictions and overlaps.
Skip if: --skills-only, --agents-only, or --freshness flag provided.
Step 1: Load Rules and Constitution
Load:
{TEAM_AI_DIRECTIVES}/context_modules/constitution.md{TEAM_AI_DIRECTIVES}/context_modules/rules/**/*.md
Step 2: Detect Conflicts
Conflict levels:
| Level | Pattern | Severity |
| Direct Contradiction | must X vs never X | CRITICAL |
| Implicit Contradiction | Numeric/logical impossibility | ERROR |
| Exception Conflict | Base rule vs exception | WARNING |
| Scope Overlap | Overlapping rules | INFO |
| Constitution Conflict | Rule vs principle | CRITICAL |
Run the conflict scan over {TEAM_AI_DIRECTIVES}/context_modules/rules inline
(no cross-skill helper — scripts stay skill-local), assigning one of:
Step 3: Create Conflict CDRs
For each conflict, create a CDR in adlc branch drafts/cdr/CDR-{NNN}.md:
## CDR-{NNN}: Resolve Rule Conflict: {title}
### Status
**Discovered**
### Date
{today}
### Source
Rule conflict detection via /team-repair --validate
### Target Module
`context_modules/rules/{domain}/`
### Context Type
Rule
### Context
**Conflict Details**:
- Rule A: {path} — "{statement}"
- Rule B: {path} — "{statement}"
- Type: {critical|error|warning|info}
### Decision
**Proposed Resolution**:
1. Add exception
2. Edit rule to avoid conflict
3. Mark intentional
4. Deprecate one ruleRegenerate the local CDR index.
Handoff: if conflict CDRs created, suggest /team-levelup.
Phase 9: Freshness Verification
Objective: Update verified timestamps for valid directives and flag stale ones.
Skip if: --skills-only, --agents-only, or --conflicts flag provided.
Step 1: Identify Valid Directives
For each context module file (rules, personas, examples, constitution) and skill SKILL.md:
- If no conflicts detected for this file → eligible for verification update
- If conflicts detected → skip (will be resolved via conflict CDRs)
Step 2: Update Verification Metadata
For each eligible directive:
- Parse YAML frontmatter
- Append to
verifiedlist:{ by: process:team-repair, at: {today}T00:00:00Z } - Update
generated.atif content changed during this repair run - Append verification entry to per-directory
log.md:
* **Verification**: Verified [{Title}](path) — no conflicts detected.Step 3: Report Stale Directives
Flag directives whose latest verified[].at is older than stale_after (default 180d), or whose status is deprecated.
### Stale Directives
| File | Last Verified | Age | Stale After | Status |
|---|---|---|---|---|
| rules/old-pattern.md | 2026-04-01 | 190d | 180d | stale |Phase 9b: Confidence Update
Skip if: --update-confidence flag is NOT provided.
Objective: Read usage data from adlc orphan branch, aggregate into
confidence scores, and update OKF frontmatter on main branch.
Step 1: Read Usage Data from adlc Branch
# Read all project usage files from adlc branch
for file in $(git -C "$TEAM_AI_DIRECTIVES" ls-tree --name-only "$ADLC_BRANCH" "reports/projects/" 2>/dev/null | grep '\.json$'); do
git -C "$TEAM_AI_DIRECTIVES" show "${ADLC_BRANCH}:${file}"
doneStep 2: Aggregate into confidence-scores.json
Merge all project JSONs, calculate:
usage_count: sum ofmatchedacross projectsapply_count: sum ofappliedacross projectssuccess_rate:apply_count / usage_countlast_used: most recentlast_usedacross projectstrend:rising(used in last 7 days),stable(last 30 days),falling(>30 days)projects: list of project names
Write to adlc branch: reports/confidence-scores.json.
Step 3: Update OKF Frontmatter on Main Branch
For each CDR in confidence-scores.json:
- Find corresponding context module file via
cdr_ref - Add/update
confidence:block in YAML frontmatter - Commit to main branch
Step 4: Rebuild CDR.md with Confidence Column
Add Confidence and Usage columns to the derived CDR.md table.
Phase 10: Build to Delete (Factor XII)
Objective: Identify directives that are no longer needed because baseline models handle them natively. This is the "Harness Decay" mechanism — run evals without directives; if the model passes independently, the directive is a candidate for removal.
Skip if: --build-to-delete flag is NOT provided.
This phase makes LLM calls — it runs goldenset cases against the agent to test whether directives are still needed.
Step 1: Load All Goldensets
Read all goldenset directories from {TEAM_AI_DIRECTIVES}/evals/:
ls -1 "$TEAM_AI_DIRECTIVES/evals/" 2>/dev/nullFor each {directive-id} directory, read:
evals/{directive-id}/goldset.md— human-readable casesevals/{directive-id}/goldset.json— machine-readable cases
If no goldensets exist, report: "No evals found — run /team-levelup to create eval CDRs first." and skip this phase.
Step 2: Identify Paired Directives
For each goldenset, identify its paired directive:
- Read
paired_directivefrom the goldenset frontmatter - Read the directive file from
context_modules/(e.g.,rules/security/sql_injection_prevention.md) - If the directive file doesn't exist → skip (already deleted or orphaned eval)
Step 3: Run Goldenset Without Directive
For each directive+eval pair:
- Temporarily remove the directive from the context that would be loaded (simulate: the agent works without the rule)
- Run the goldenset cases against the agent via LLM calls:
- For each pass case: present the scenario and input context, ask the agent to produce output, check if it follows the (removed) directive
- For each fail case: present the scenario and input context, ask the agent to produce output, check if it still makes the mistake
- Compute pass rate:
cases_passed / total_cases
Step 4: Classify Results
| Pass Rate | Verdict | Recommendation |
| 100% | Delete candidate | Model handles this natively — directive is obsolete |
| 80-99% | Review candidate | Model mostly handles it — consider simplifying the directive |
| < 80% | Keep | Model still needs the directive |
For every Keep (and Review candidate) verdict, ask the complementary promote-to-check question (EVAL-010): can a deterministic check (unit test / binary grader / pre-commit hook / lint rule / CI job) mechanically enforce this rule? If yes, it is a Promotion candidate — pay once for a check instead of re-injecting a fuzzy rule into every session.
Step 5: Generate Harness Decay Report
## Build to Delete Report
### Candidates for Removal (model passes 100% without directive)
| Directive | Eval | Pass Rate | Recommendation |
|---|---|---|---|
| rules/security/sql_injection.md | evals/CDR-001/ | 100% (6/6) | Delete — model handles this natively now |
### Review Candidates (80-99%)
| Directive | Eval | Pass Rate | Recommendation |
|---|---|---|---|
| rules/devops/helm_packaging.md | evals/CDR-008/ | 83% (5/6) | Simplify — model mostly handles it, 1 case failed |
### Still Needed (< 80%)
| Directive | Eval | Pass Rate | Recommendation |
|---|---|---|---|
| rules/style/python_pep8.md | evals/CDR-015/ | 40% (2/5) | Keep — model still needs guidance |
### Promotion Candidates (Keep, but check-enforceable)
| Directive | Eval | Pass Rate | Proposed Check |
|---|---|---|---|
| rules/architecture/import_boundaries.md | evals/CDR-011/ | 55% (3/5) | Promote to pre-commit lint rule — file-location pattern is mechanical |Step 6: Create Deletion CDRs (and Promotion CDRs)
For each Delete candidate (100% pass rate), create a CDR in adlc branch drafts/cdr/CDR-{NNN}.md:
## CDR-{NNN}: Delete Directive: [Title]
### Status: **Discovered**
### Date: [YYYY-MM-DD]
### Source: Build to Delete via /team-repair --build-to-delete
### Target Module: `context_modules/rules/{domain}/{file}.md` + `evals/{directive-id}/`
### Context Type: Constitution Amendment
### Descriptor: Directive is obsolete — model handles natively without the rule.
### Context
The directive `{title}` was tested by running its goldenset cases without the directive loaded.
The model passed 100% of cases (N/N), indicating the baseline model now handles this pattern natively.
The directive is a candidate for Harness Decay removal.
### Decision
Delete both the directive file and its paired eval goldenset.
### Evidence
- Directive: context_modules/rules/{domain}/{file}.md
- Eval: evals/{directive-id}/goldset.md
- Pass rate: 100% (N/N cases passed without the directive)
- Test date: [YYYY-MM-DD]Regenerate the local CDR index. Handoff: suggest /team-levelup to review deletion candidates.
For each Promotion candidate, create a CDR in adlc branch drafts/cdr/CDR-{NNN}.md:
## CDR-{NNN}: Promote Directive to Deterministic Check: [Title]
### Status: **Discovered**
### Date: [YYYY-MM-DD]
### Source: Build to Delete via /team-repair --build-to-delete
### Target Module: `context_modules/rules/{domain}/{file}.md`
### Context Type: Rule
### Descriptor: Rule is mechanically enforceable — promote to a deterministic check.
### Context
The directive `{title}` survived build-to-delete (model still needs it, pass rate < 100%),
but its pattern is mechanical — a deterministic check (unit test / binary grader /
pre-commit hook / lint rule / CI job) can enforce it without session context.
### Decision
Build the deterministic check. Once it exists and runs in CI, deprecate the CDR
or reduce it to a thin pointer (`enforced by <check path>`). Route to
`/team-levelup` action **P — Promote to check** (Phase 2b).
### Evidence
- Directive: context_modules/rules/{domain}/{file}.md
- Proposed check vehicle: [unit test | grader | pre-commit | lint | CI job]
- Pass rate without directive: N% (M/K cases) — rule still needed
- Test date: [YYYY-MM-DD]Regenerate the local CDR index again. Handoff: suggest /team-levelup to review promotion candidates (action P).
Phase 11: Validate Drafts
Objective: Validate draft decision records in .adlc/drafts/ for structural completeness — required frontmatter fields, required body sections, and valid enum values. This is a read-only validation mode — no files are created, modified, or deleted.
Skip if: --validate-drafts flag is NOT provided.
This phase makes NO LLM calls — it is purely mechanical file parsing and validation, like a linter.
Step 1: Scan Draft Directories
Scan all files in the following draft directories relative to {REPO_ROOT}:
.adlc/drafts/adr/
.adlc/drafts/pdr/
.adlc/drafts/chdr/
adlc branch drafts/cdr/
.adlc/drafts/evals/For each directory, list all *.md files. If a directory does not exist, skip it silently (not an error — that draft type simply has no drafts).
If no draft files are found in any directory, report: "No draft files found in .adlc/drafts/ — nothing to validate." and skip this phase.
Step 2: Validate Frontmatter
For each draft file, parse YAML frontmatter (between --- delimiters) and verify the following required fields are present:
| Field | Required | Valid Values | Notes |
status | Yes | proposed, accepted, rejected, deferred, superseded, discovered | Case-insensitive match |
date | Yes | Any non-empty string (expected YYYY-MM-DD) | Must not be empty or placeholder |
type | Yes | decision, product, pattern, incident, workaround, constraint, abandoned, eval | Case-insensitive match |
evidence | Yes | confirmed, inferred, unknown | Case-insensitive match |
source | Yes | Any non-empty string | Origin of the draft (skill name, session, etc.) |
revisit-when | Yes (present) | Empty string, N/A, or any non-empty string | Field must exist; value can be empty or N/A |
For each missing or invalid field, record:
- File path
- Line number (of the frontmatter key, or line 1 if frontmatter is entirely missing)
- Issue description (e.g., "Missing required field: status", "Invalid status value: 'draft' — expected one of: proposed, accepted, rejected, deferred, superseded, discovered")
Step 3: Validate Body Sections
After frontmatter, validate that the following required body sections are present as Markdown headings:
| Section | Required | Notes |
## Context (or ### Context) | Yes | Must exist as a heading |
## Decision (or ### Decision) | Yes | Must exist as a heading |
## Rejected Alternatives (or ### Rejected Alternatives) | Yes | Must exist as a heading |
## Reason (or ### Reason) | Yes | Must exist as a heading |
Heading level flexibility: accept ## or ### (or even ####) for each section. Match by heading text (case-insensitive, trimmed).
For each missing section, record:
- File path
- Line number (0 if section not found — report as "section not found")
- Issue description (e.g., "Missing required body section: Rejected Alternatives")
Step 4: Conditional Validation — Rejected Alternatives Non-Empty
For draft files where type is decision or abandoned, the ## Rejected Alternatives (or ### Rejected Alternatives) section must not be empty. "Empty" means:
- No content between the heading and the next heading or end of file
- Only whitespace or placeholder text (e.g., "N/A", "TBD", "TODO", "none")
For each violation, record:
- File path
- Line number of the heading
- Issue: "Rejected Alternatives section is empty for decision/abandoned type — must list at least one rejected alternative"
Step 5: Generate Validation Report
## Draft Validation Report
**Date**: {date}
**Mode**: VALIDATE ONLY (no modifications)
### Summary
| Metric | Count |
|--------|-------|
| Draft directories scanned | {n} |
| Draft files validated | {n} |
| Files with errors | {n} |
| Files with warnings | {n} |
| Files passing validation | {n} |
| Total findings | {n} |
### Findings
| Severity | File | Line | Issue |
|----------|------|------|-------|
| Error | .adlc/drafts/adr/ADR-001.md | 3 | Missing required field: evidence |
| Error | .adlc/drafts/adr/ADR-001.md | 0 | Missing required body section: Reason |
| Error | .adlc/drafts/pdr/PDR-005.md | 7 | Invalid status value: 'draft' — expected one of: proposed, accepted, rejected, deferred, superseded, discovered |
| Warning | adlc branch drafts/cdr/CDR-010.md | 15 | Rejected Alternatives section is empty for decision type — must list at least one rejected alternative |
{If no findings:}
> **All draft files passed validation — no issues found.**Step 6: Handoff
- If errors were found: suggest fixing the draft files before promoting them via the appropriate clarify skill (
/architect-clarifyfor ADRs,/product-clarifyfor PDRs,/change-clarifyfor ChDRs,/team-levelupfor CDRs). - If all drafts pass validation: confirm drafts are structurally ready for promotion.
- Remind: validation does not check semantic quality — only structural completeness. A draft that passes validation may still be rejected during clarification.
Phase 12: Summary Report
## Team Repair Summary
**Date**: {date}
**Team Directives**: {path}
**Mode**: {DRY RUN|LIVE}
### AGENTS.md Repair
| Status | Action |
|--------|--------|
| {VALID|CREATED|OVERWRITTEN} | {No changes needed|Created from template|Re-created from template} |
### OKF index.md + log.md + CDR.md Repair
| Action | Count |
|--------|-------|
| Files scanned | {n} |
| v0.1→v0.2 migrated | {n} |
| index.md files rebuilt | {n} |
| log.md files rebuilt | {n} |
| CDR.md derived entries | {n} |
| Orphans repaired | {n} |
| Missing removed | {n} |
### .skills.json Repair
| Action | Count |
|--------|-------|
| Skills scanned | {n} |
| Valid entries | {n} |
| Orphans repaired | {n} |
| Missing removed | {n} |
### Conflict Scanning
| Metric | Count |
|---|---|
| Conflicts detected | {n} |
| Conflict CDRs created | {n} |
| Critical | {n} |
| Error | {n} |
| Warning | {n} |
| Info | {n} |
### Freshness Verification
| Metric | Count |
|---|---|---|
| Directives updated | {n} |
| Stale directives (>30d) | {n} |
| Skipped (has conflicts) | {n} |
### Draft Validation
| Metric | Count |
|--------|-------|
| Draft files validated | {n} |
| Files with errors | {n} |
| Files with warnings | {n} |
| Files passing validation | {n} |
| Total findings | {n} |
{If --validate-drafts was not run:}
> **Note**: Draft validation not run (use `--validate-drafts` to validate .adlc/drafts/)
### Files Modified
| File | Change |
|------|--------|
| {file} | {change description} |
{If --dry-run:}
> **Note**: Dry run mode - no files were modified
### Next Steps
1. Review repaired files
2. If conflict CDRs were created, run `/team-levelup` to resolve them
3. Commit changes if satisfiedNotes
- Auto-fix: Always repairs issues automatically (no confirmation needed)
- Dry run: Use
--dry-runto preview changes without writing - Selective repair: Use
--index-only,--skills-only, or--agents-onlyfor specific targets - Validation modes:
--validateruns conflict scan + freshness;--conflictsand--freshnessrun each separately - Draft validation:
--validate-draftsvalidates.adlc/drafts/{adr,pdr,chdr,cdr,evals}/— read-only, no modifications (like--build-to-delete, it only reports findings) - YAML frontmatter: Auto-generated for orphan context modules
- Skills entries: Auto-generated from SKILL.md content
- AGENTS.md: Overwrites if corrupted (missing required sections)
- Idempotent: Re-running produces same result
Common Rationalizations
| Rationalization | Reality |
| "The indexes look fine — no need to reindex." | Orphaned files and missing frontmatter are invisible without a full directory scan. |
| "I'll just hand-edit CDR.md to add the missing row." | Manual edits drift from actual content; a rebuild guarantees the index matches the filesystem. |
| "Dry run is unnecessary — just write the changes." | A dry run surfaces unexpected orphans and null CDR refs before any file is mutated. |
| "AGENTS.md looks valid, so I'll skip Phase 2." | Missing sections can be subtle (e.g., a renamed heading). Validation is cheap and idempotent. |
| "Skipping Step 5 — the project AGENTS.md is not my job." | The team AI directives' own AGENTS.md describes structure; the project-level AGENTS.md is what tells agents to invoke team-boot at session start. Without it, the directives remain invisible. |
| "I can skip the CDR_LOOKUP step for orphans." | Without the lookup, existing CDR refs are lost and orphaned entries get cdr_ref: null, breaking traceability. |
| "I'll just jump to the repair — no need for a health check first." | Phase 0 exists precisely because an unhealthy framework makes repairs dangerous or meaningless. Run it. |
"A [WARN] on Phase 0 is basically an [OK]." | Warnings are non-blocking for exit code but often signal drift that becomes a [FAIL] later. Track warnings across runs. |
Red Flags
- Overwriting AGENTS.md without validating structure first — a "corrupted" verdict should require evidence of missing sections, not a hunch; otherwise custom content is destroyed.
- Generating
cdr_ref: nullwhen an existing CDR_LOOKUP entry exists — this silently severs the audit trail between a context module and its accepted CDR record. - Skipping the dry run when the orphan count is high — bulk auto-fix without review leads to fabricated IDs and metadata propagating into version control.
- Writing
.skills.jsonentries without parsing the actualSKILL.md— fabricated descriptions and categories make skills unsearchable and misrepresent capabilities. - Proceeding past Phase 2 when
TEAM_AI_DIRECTIVESis empty — operating without a configured repository writes to undefined paths and corrupts the wrong workspace. - Skipping Phase 0 Health Check — jumping straight into repairs without verifying the framework is installed risks writing to an absent or misconfigured workspace.
- Skipping Step 5 (project AGENTS.md injection) — the team AI directives' own
AGENTS.mddescribes its structure, but the project-levelAGENTS.mdis what tells agents to invoketeam-bootat session start. Without it, agents have no session-start instruction and the team AI directives remains invisible until manually loaded.
Verification
- Phase 0 Health Check passes all 8 checks (no
[FAIL]) before any repair is attempted. - AGENTS.md exists at
{TEAM_AI_DIRECTIVES}/AGENTS.mdand contains all six required sections. - Project-level
AGENTS.mdat{REPO_ROOT}/AGENTS.mdcontains the<!-- TEAM_AI_DIRECTIVES START -->managed section with the event-hook awareness note, fallbackteam-bootinvocation, Class Boots catalog, Team Context in Use output contract, and compact Decision Capture triggers. - CDR.md entry count equals the number of scanned context module
.mdfiles (excludingconstitution.md). - Every context module file under
context_modules/{rules,personas,examples}/has YAML frontmatter with a non-emptyidfield. - Every
cdr_refin orphan frontmatter matches the pre-existing CDR lookup (no regression tonullwhere a prior ref existed). - Every skill directory containing a
SKILL.mdhas a corresponding entry in.skills.json. - No
.skills.jsonentry references a skill directory that does not exist on disk. - The summary report lists non-zero counts for "Files scanned" / "Skills scanned" and shows consistent totals.
- Re-running the skill with no flags produces zero "Files Modified" entries (idempotency check).
- Conflict scan completed (if not skipped) and conflict CDRs created for any findings.
- Freshness verification completed (if not skipped) and stale directives reported.
- No rule contradictions remain unreported after
--validate. - If
--validate-draftswas run: every draft file in.adlc/drafts/{adr,pdr,chdr,cdr,evals}/was scanned and findings reported (if any). - If
--validate-draftswas run: no files in.adlc/drafts/were created, modified, or deleted (validation-only, read-only mode). - If
--validate-draftswas run: draft validation findings include file path, line number, and issue description for each finding.
Configuration
TEAM_AI_DIRECTIVES— Path to the team AI directives (overrides.adlc/init-options.json)..adlc/init-options.json— Project-level config file withteam_ai_directivesfield.- Default fallback:
team-ai-directives/relative to project root. scripts/bash/setup-team.sh/scripts/powershell/setup-team.ps1— Skill-local path resolution (--json/-Json).
12-Factor Alignment
Factor XI (Directives as Code) — maintains integrity of version-controlled team directives.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

