agentleFS
Sign inStart free
myshra777-aimyshra777-ai/garuda/skills/pre-edit-governance/SKILL.md

pre-edit-governance

Skill4 starsChanged 13 days agoApache-2.0View on GitHub

Use before an AI coding agent modifies code, or whenever someone asks whether a planned change is safe to make. This is the mandatory checkpoint that runs first: in one call Garuda returns the target's caller fan-in, its criticality tier, the active YAML policies that apply, and the MUST and MUST NOT rules from ADRs and agent instruction files. It also flags metrics or calculations with competing implementations across teams and any surface frozen against unreviewed edits. The result is a proceed, revise or escalate decision, so the agent knows what to respect, or when to ask a human, before writing any code. Read-only by default.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    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
  1. Very specific:

    Epistemic States: Always distinguish SUPPORTED (backed by compiler/AST + runtime evidence where required), UNVERIFIED (statically …

    line 35

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

03 — Garuda Pre-Edit Governance Gate

Acts as the mandatory pre-change checkpoint before an AI coding agent modifies code, returning a proceed, revise, or escalate decision.

1. Purpose & Triggers

Invoke this skill when the user or coding workflow asks:

  • "Before modifying this code, check Garuda's policies"
  • "Run pre-edit governance for my planned change"
  • "Can an AI agent safely edit this package?"

2. Required Context

  • workspace: Target Garuda workspace name (default: "default").
  • repo_dir: Repository root path (default: ".").
  • diff / target: Git revision range (e.g. "HEAD") or symbol/package name.

3. Ordered MCP Workflow

  1. garuda.preflight_bundle (or garuda.talon_pre_edit) — Fetch caller fan-in, package criticality tier, active YAML policies, and applicable MUST / MUST_NOT ADR claims in one call.
  2. garuda.disambiguate_logic — If the target symbol implements a domain metric or calculation, check for competing cross-team variants (0.0–1.0 weights) and FROZEN_SEVERE_DO_NOT_TOUCH locks.
  3. Evaluate whether the target surface is HIGH_SEVERE or governed by a BLOCK policy before writing code.

CLI Fallback (when MCP stdio is not attached):

garuda talon brief <SymbolOrFile> && garuda audit criticality

4. Decision & Epistemic Rules

  • Epistemic States: Always distinguish SUPPORTED (backed by compiler/AST + runtime evidence where required), UNVERIFIED (statically extracted, awaiting runtime trace), and CONTRADICTED (violates ADR, policy, or static boundary).
  • Governance Verdicts: ALLOW (proceed), WARN (proceed with note), REVIEW (requires human architectural review), BLOCK (halt and remediate net-new violation), INCONCLUSIVE (0 entities matched policy scope; never treat as vacuous ALLOW).

5. Output Contract

Format every report using this six-part structure: Finding -> Evidence (file:line & match_method) -> Affected Components (callers & tier) -> Policy / ADR -> Confidence / Status -> Recommended Action

6. Safety & Limitations

  • Read-Only by Default: Public installations run with GARUDA_MCP_READ_ONLY=true. Never invoke mutating tools (garuda.approve_base_rulebook, garuda.decide_logic_variant, garuda.reconcile_data_audit) without explicit human authorization (GARUDA_MCP_APPROVER).
  • Scope Qualification: RFC 6962 Merkle proofs (H_leaf = SHA256(0x00 || data)) establish tamper-evident integrity and inclusion of recorded governance evaluations; they do not replace unit, integration, or dynamic security testing.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free