fabric-heal
Skill49 starsChanged 6 months agoMITView on GitHub
Self-healing skill that improves signal mapper keyword coverage and capability coverage (ADR-0002) through iterative problem generation and keyword/capability patching. Use when user says "heal signal mapper", "improve keyword coverage", "improve capability coverage", "generate problem statements", "run healing loop", "patch signal mapper", or asks about "signal mapper gaps" or "capability gaps". Do NOT use for project architecture or deployment.
In this repository8 other files it gives its agents.
- .github/copilot-instructions.mdCopilot instructions
- fabric-deploySkill
- fabric-designSkill
- fabric-discoverSkill
- fabric-documentSkill
- fabric-testSkill
- grill-with-docsSkill
- handoffSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 37
Use 5-8 categories per batch, 3-5 problems per category.
- Very specific: line 44
Update _shared/registry/signal-categories.json using signal-categories-cli.py (max 15 new keywords per category per iteration)
- Very specific: line 63
Use signal-mapper.py for all signal lookups.
- Distinctive: line 65
Never remove existing keywords or modify the matching algorithm
- Distinctive: line 67
Prefer specific terms over generic ones
≈ “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
Fabric Signal Mapper Healer
Mode 1: Generate Problem Statements
Orchestrator passes --mode generate --batch-num N. Write 25 novel problem statements to problem-statements.md.
Requirements
Realistic (enterprise user), specific (volumes/tools/deadlines), multi-signal (2-3 categories), domain-diverse, varying complexity.
Output Format (parser-dependent)
# Problem Statements for Stress Testing
> Auto-generated batch N — 25 problems for self-healing loop.
## Category Name
1. "Problem statement text in quotes."
2. "Another problem in same category."Use 5-8 categories per batch, 3-5 problems per category.
Mode 2: Analyze Gaps and Patch Keywords
Orchestrator passes --mode heal with benchmark results (coverage %, zero-candidate count, uncovered terms).
- Map uncovered terms to signal categories (1-11)
- Update
_shared/registry/signal-categories.jsonusingsignal-categories-cli.py(max 15 new keywords per category per iteration)
Orchestrator Reporting (Verbose Distribution)
heal-orchestrator.py prints a verbose, deterministic distribution report with ASCII bar charts:
- Task flow suggestion rate
- Top candidate distribution
- Pattern-category distribution (batch/streaming/hybrid/governance/app-api)
- Skillset distribution (LC/CF/mixed tendencies)
- Item type frequency across suggested flows
- Recommendations derived from benchmark thresholds and recurring uncovered terms
Detailed analytics are persisted to:
.github/skills/fabric-heal/scripts/_heal-loop-results.json
Constraints
- Use
signal-mapper.pyfor all signal lookups. - For keyword maintenance, use
_shared/scripts/signal-categories-cli.py(add/remove/move/list) instead of manual raw JSON edits. - Never remove existing keywords or modify the matching algorithm
- Clean up generated files: delete
problem-statements-batch*.mdafter each loop completes - Prefer specific terms over generic ones
References
.github/skills/fabric-discover/scripts/signal-mapper.py— current keyword categories_shared/scripts/signal-categories-cli.py— safe keyword maintenance helper for signal categories registryproblem-statements.md— existing format (in this skill's folder)task-flows.md— task flow descriptions
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.

