engineering-governance
Skill6 starsChanged 19 days agoMITView on GitHub
Apply human decision gates, milestone review, physical-node routing and bounded worker repairs in an authorized multi-agent engineering project. Use for costly jobs, cross-host handoffs, disputed scope or independent result verification; avoid imposing the full profile on routine single-agent fixes.
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- claim-ledgerSkill
- critic-reviewSkill
- distributed-orchestrationSkill
- engineering-planSkill
- execution-integritySkill
- literature-provenanceSkill
- multi-agent-collaborationSkill
- technical-writingSkill
- worker-taskSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Read the reference relevant to the task rather than loading the whole suite:
- Distinctive: line 18
Before declaring a governance improvement, show the observable evidence.
≈ “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.
Files
Apply the project's adopted governance profile within the user's authorization and host instructions. H is the human principal and highest project decision authority, never a model or simulated role. Human decisions override agent consensus and project procedure; they do not manufacture evidence or bypass host constraints.
Read the reference relevant to the task rather than loading the whole suite:
- Scope deadlock, budget stop or decision override: human-gate.md.
- Expensive runs, idle token cost or review scheduling: milestone-mode.md.
- Missing hardware, entitlement, network capability or cross-host migration: node-affinity.md.
- Code ownership, worker self-fixes or blind verification: role-boundaries.md.
- Administrative stalls, scoped autonomy or merged roles: lean execution.
- Watcher recovery, quota waits or unproven delivery: operational health.
Use H/O/F/W/P/C/V/A for the governed profile. I is a compatibility alias for F in the basic walkthrough, not an extra writer. Preserve existing project roles unless the project adopts this profile. A skill does not start agents, enable watchers or expand permissions.
Before declaring a governance improvement, show the observable evidence. Token-reduction percentages require matched workload measurements; a new protocol or successful launch is not a verified outcome.
The role/write limits are starter defaults. A project's explicitly adopted lean profile may assign bounded implementation or operational repairs to another role, with clear file ownership, invariants and task-level limits. Do not require a new acknowledgment for already-authorized ordinary work. Retain prospective scope/resource decisions and independent outcome checking. For expensive staged work, execution-integrity supplies exact-byte packets and durable attempt reservations; a preparation packet is never execution permission.
For cross-provider/physical-node execution, the optional distributed-orchestration skill supplies capability-based assignment and safe tail planning. For source evidence, literature-provenance supplies registry citation checks and exact offline quote location. These are complementary tools, not permission to acquire restricted sources or trust one provider's outputs. Registered runs must freeze executable criteria/tests before data and preserve failed verdicts; toolkit preregistration utilities require a protected anchor and serialized ledger authority.
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.

