propose-workflow-improvements
Skill2 starsChanged 7 months agoMITView on GitHub
Convert Dream workflow-optimization candidates into evidence-backed Escrow proposals using a fixed Observer/Analyzer split.
In this repository26 other files it gives its agents.
- advise-tech-stackSkill
- commit-with-affectsSkill
- db-dev-guideSkill
- export-db-indexesSkill
- frontend-dev-guideSkill
- frontend-ux-evaluatorSkill
- generate-pr-descriptionSkill
- generate-test-from-impactSkill
- generate-test-skeletonSkill
- lock-global-conventionsSkill
- plan-component-hierarchySkill
- plan-page-mapSkill
- proposal-reviewSkill
- pr-self-reviewSkill
- record-decisionSkill
- review-guardrailsSkill
- run-pending-specsSkill
- run-testsSkill
- scan-bundle-bloatSkill
- scan-code-hygieneSkill
- scan-dead-routesSkill
- scan-frontend-qualitySkill
- scan-orphan-componentsSkill
- scan-unused-exportsSkill
- sync-llm-contextSkill
- verify-specSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
Read project.config.json and use:
- Distinctive: line 26
Before running, check for an Analyzer freeze:
≈ “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
Propose Workflow Improvements
Architecture
| Phase | Responsibility | Mutability |
| Observer | Collect and normalize raw signals | Stable; change only by direct human edit |
| Analyzer | Group patterns, filter by constitution, write Escrow proposals | Evolves through reviewed workflow changes |
Analyzer reads only the Observer snapshot. It must not directly reinterpret raw files after the snapshot is written.
Paths
Read project.config.json and use:
agent.signal_dirdefaulttmp/agent-signalsagent.management_dirdefaultdocs/agent-runs/managementagent.delivery_dirdefaultdocs/agent-runs/delivery
Before running, check for an Analyzer freeze:
test -f {management_dir}/analyzer-freeze.md \
&& ! rg -q "manual review complete, resume" {management_dir}/analyzer-freeze.md \
&& echo "[FROZEN] Analyzer is frozen until manual review resumes it"Phase 1: Observer Snapshot
Collect:
- Latest
{management_dir}/workflow-optimization-candidates-*.md - Historical candidates under
{management_dir}and{delivery_dir} - Historical proposals and Escrow files
{management_dir}/proposals-status.mdif present- Matching
session-*.mdandprereview-*.mdevidence fromagent.signal_dir
Write:
## Observer Snapshot - {timestamp}
### Current Candidates
{latest candidates content}
### Historical Candidate Summary
- {title} - first {date}, latest {date}, count {n}
### Historical Proposal Summary
- {title} - status {implemented/not implemented/partial}, effect {resolved/unresolved/unknown}
### Evidence Summary
- {slug}: session {path}, prereview {path or none}Phase 2: Analyzer
Use .agents/rules/propose-wi-constitution.md.
For each candidate:
- Assign one impact layer:
spec-format,skill-logic,toolchain, orflow-handoff. - Confirm direct evidence exists in a session or prereview file.
- Group with historical candidates using the constitution's similarity rule.
- Mark as first occurrence, recurrence, or likely resolved.
- Drop anything that violates the constitution.
Output: Escrow Proposal
Write {management_dir}/proposals-escrow-{YYYYMMDD}.md:
# Workflow Improvement Proposals (Escrow) - {date}
**Status**: pending Dream confirmation before becoming Actionable
**Based on**: workflow-optimization-candidates-{date}.md
### Proposal N: {title}
**Impact layer**: {spec-format / skill-logic / toolchain / flow-handoff}
**History**: first occurrence / recurrence / likely resolved
**Evidence**: {session/prereview paths}
**Observed problem**: {specific behavior}
**Root cause**: {why the workflow allowed it}
**Suggested change**:
- File: {path}
- Change: {paragraph-level edit}
**Expected effect**: {what stops recurring}
**Review focus**: {what a human should verify}
## Filtered By Constitution
- {title}: violates I-{n}; {reason}
## Manual Review Needed
- {title}: suspected Observer change; human decision required
## Low Priority
- {title}: likely resolved; continue watchingAppend a confidence report with counts for total candidates, Escrow entries, filtered entries, and unresolved Actionable items.
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.

