agency-invocation
Skill1 starsChanged 56 days agoMITView on GitHub
Invoke the SDD agency: /agency and per-stage bundles.
In this repository67 other files it gives its agents.
- architect-reviewerSkill
- backend-architectSkill
- backend-developerSkill
- code-architectSkill
- codebase-explorerSkill
- code-reviewerSkill
- code-simplifierSkill
- debuggerSkill
- error-detectiveSkill
- fullstack-developerSkill
- git-workflow-managerSkill
- legacy-modernizerSkill
- pragmatic-architectSkill
- prdSkill
- qa-expertSkill
- research-technical-spikeSkill
- sdd-spec-writerSkill
- security-auditorSkill
- supply-chain-securitySkill
- task-decomposition-expertSkill
- technical-debt-managerSkill
- test-engineerSkill
- typescript-proSkill
- architecture-decision-recordsSkill
- hermes-sdd-packagingSkill
- bug-diagnosisSkill
- build-craftSkill
- code-healthSkill
- code-review-checklistSkill
- commit-smartSkill
- context-architectureSkill
- design-explorationSkill
- design-to-htmlSkill
- developer-experience-reviewSkill
- diagram-tripletSkill
- dispatching-parallel-agentsSkill
- document-diataxisSkill
- e2e-testing-patternsSkill
- executing-plansSkill
- module-designSkill
- openspec-apply-changeSkill
- openspec-archive-changeSkill
- openspec-bulk-archive-changeSkill
- openspec-continue-changeSkill
- openspec-exploreSkill
- openspec-ff-changeSkill
- openspec-new-changeSkill
- openspec-onboardSkill
- openspec-proposeSkill
- openspec-sync-specsSkill
- openspec-update-changeSkill
- openspec-verify-changeSkill
- plan-scope-reviewSkill
- project-learningsSkill
- qa-test-plannerSkill
- requirements-claritySkill
- review-structuralSkill
- security-evidence-firstSkill
- hermes-sdd-orchestrationSkill
- openspec-sddSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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 16
All of them are skill bundles in ~/.hermes/skill-bundles/.
- Distinctive: line 36
No implementation code without a validated OpenSpec change.
- Distinctive: line 39
Only review and QA can block.
- Distinctive: line 41
Every closure carries a final report, even a blocked one.
- Distinctive: line 46
Every reply to the user is in Spanish — including one-line acknowledgements and closing summaries after an English-only task.
- Distinctive: line 54
Before sending a reply, check the prose itself — tool output, code blocks, paths, slugs and commands stay in English by design and do not …
≈ “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
Agency invocation (cheat sheet)
How the Hermes SDD agency is used. Full detail: ~/.hermes/HOW-TO-INVOKE-THE-AGENCY.md.
Slash commands
/agency— general orchestrator; asks for the stage if you did not name one.- Per stage:
/init-project·/idea·/architecture·/plan·/implement·/review·/qa·/release
All of them are skill bundles in ~/.hermes/skill-bundles/; each loads
hermes-sdd-orchestration + openspec-sdd and injects the stage protocol.
hermes bundles list to see them · hermes bundles show <n> to see what each loads.
Without slash commands
The hermes-sdd-orchestration skill autoloads on: "the agency", "los agentes",
"run the SDD loop", "initialize the project", "the workflow".
Before you start
cd <project-root> && pwd -P && openspec context --json
test -f openspec/project.md || echo "MISSING → /init-project"Ambiguous root ⇒ the agency stops and reports. It never proceeds with the "most likely" one.
Hard rules
- No implementation code without a validated OpenSpec change.
- OpenSpec lives in the project repo, never in
~/.hermes/. - Agents never talk to each other; Hermes validates every output in the real repo.
- Only review and QA can block.
- Business/architecture/security/scope ambiguity ⇒ the human decides.
- Every closure carries a final report, even a blocked one.
Output language (verify before sending)
The agency's files, agents, workflows, rules, templates and specs are all in English. Every reply to the user is in Spanish — including one-line acknowledgements and closing summaries after an English-only task.
This has failed in practice: a full Spanish verification report (translation of 33 files) came back written in Chinese, while the underlying tool output was correct. The language of the work and the language of the answer are independent decisions — a task conducted in English does not license answering in English, and never in a third language.
Before sending a reply, check the prose itself — tool output, code blocks, paths, slugs and commands stay in English by design and do not count. If a reply goes out in the wrong language, say so plainly in the next turn (the sent message cannot be recalled) and restate the content in Spanish.
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.

