techtide-harness-kit
AGENTS.md0 starsChanged 6 months agoApache-2.0View on GitHub
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository10 other files it gives its agents.
- agents/AGENTS.mdAGENTS.md
- catalog/AGENTS.mdAGENTS.md
- mcp/AGENTS.mdAGENTS.md
- rules/AGENTS.mdAGENTS.md
- schemas/AGENTS.mdAGENTS.md
- skills/AGENTS.mdAGENTS.md
- templates/AGENTS.mdAGENTS.md
- tests/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
The file
We split this AGENTS.md into the rules it gives an agent
9 rules. About 1 is common, 4 are distinctive, and 4 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 36
Update catalog JSON when adding, moving, or removing cataloged assets.
- Very specific: line 38
Keep skill frontmatter metadata under metadata, including metadata.version and metadata.author.
- Very specific: line 39
Keep author and version truth in the canonical contract plus adjacent metadata.json unless a harness's official docs verify executable …
- Distinctive: line 40
Keep README human-friendly.
- Common: line 41
Do not add secrets, credentials, tokens, wallets, tenant IDs, or customer data.
- Distinctive: line 42
Prefer official docs and live evidence over memory for cloud/compliance claims.
- Distinctive: line 43
Treat broad permissions, destructive automation, and MCP mutation paths as high-risk.
- Very specific: line 44
When adding agents, update catalog/install-roles.json if the agent belongs to one or more roles.
- Very specific: line 45
All live-guard and review agents must produce the evidence fields defined in docs/evidence-output-spec.md.
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md 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
AGENTS.md
Purpose
- Cloud and zero-trust agentic marketplace for skills, agents, rules, MCP references, and npm distribution.
- Optimize for evidence-backed security workflows: least privilege, source grounding, manifests, validation, and safe automation.
Stack Map
.code-review-graph/-> generated local graph cache; do not edit..git/-> Git internals; do not edit.agents/-> Markdown/JSON agent definitions; provider/domain layout.assets/-> curated logos and visual assets.catalog/-> JSON marketplace indexes, skill integrity manifest, and role taxonomy.docs/-> lean production docs for install, compatibility, taxonomy, release, security, and quality guidance.mcp/-> Markdown/JSON MCP references.rules/-> Markdown/JSON harness rules.schemas/-> JSON Schema metadata contracts.skills/-> Markdown/JSON skill packages with reference files.templates/-> starter Markdown/JSON asset templates.tests/-> validation scripts.package.json-> npm package metadata and validation scripts.
Workflows
npm run validate-> full production validation: catalog, schemas, links, integrity, trust matrix, install coverage, routing, plugin manifests, Kiro Powers, and marketplace checks.npm run lint:docs-> advisory markdownlint + codespell.npm run manifest:write-> refreshcatalog/skill-manifest.jsonafter intentional skill edits.npm run plugin-manifest:write-> regenerate.claude-plugin/plugin.jsonfromcatalog/agents.jsonafter intentional agent additions or removals.npm run cursor-plugin:write-> regenerate.cursor-plugin/plugin.jsonfromcatalog/agents.json.npm run kiro-powers:write-> regenerate Kiro Powers underpowers/techtide-*.npm run asset-integrity:write-> refreshcatalog/asset-integrity.jsonafter intentional file additions/removals.npm pack --dry-run-> inspect npm package contents before publish.thk-export-agents --list-roles-> list available role IDs with agent counts.thk-export-agents --platform <platform> --role <role-id> --repo <path>-> install agents for a role.thk-export-agents --platform <platform> --role <role-id> --provider <provider> --repo <path>-> install role agents for one provider.
Change Rules
- Update catalog JSON when adding, moving, or removing cataloged assets.
- Regenerate the skill manifest after any intentional change under cataloged
skills/**. - Keep skill frontmatter metadata under
metadata, includingmetadata.versionandmetadata.author; use the GitHub-style author value such asgithub: TechTide. - Keep
authorandversiontruth in the canonical contract plus adjacentmetadata.jsonunless a harness's official docs verify executable metadata support. - Keep README human-friendly; keep this file agent-focused and compressed.
- Do not add secrets, credentials, tokens, wallets, tenant IDs, or customer data.
- Prefer official docs and live evidence over memory for cloud/compliance claims.
- Treat broad permissions, destructive automation, and MCP mutation paths as high-risk.
- When adding agents, update
catalog/install-roles.jsonif the agent belongs to one or more roles. - All live-guard and review agents must produce the evidence fields defined in
docs/evidence-output-spec.md.
Role-Based Pattern
catalog/install-roles.json defines six cross-provider roles. Each role is a curated list of agent and skill IDs that practitioners in that function need.
| Role ID | Who uses it |
cloud-security-engineer | IAM reviewers, security posture teams, compliance engineers |
cloud-platform-engineer | Infrastructure/SRE, IaC owners, Kubernetes platform teams |
cloud-dba | Database administrators, data platform engineers |
cloud-finops-analyst | FinOps leads, cost governance teams |
cloud-solutions-architect | Cloud architects, migration leads, generative platform engineers |
cloud-devops-engineer | CI/CD engineers, release managers, SRE ops |
Roles overlap intentionally; an agent useful to both a security engineer and a platform engineer appears in both lists.
Pipeline enforcement is documented in docs/ci-cd-enforcement-pattern.md.
Evidence output for structured verdict responses is documented in docs/evidence-output-spec.md.
Load When
- editing
agents/->agents/AGENTS.md - editing
catalog/->catalog/AGENTS.md - editing
docs/-> rootAGENTS.md - editing
mcp/->mcp/AGENTS.md - editing
rules/->rules/AGENTS.md - editing
schemas/->schemas/AGENTS.md - editing
skills/->skills/AGENTS.md - editing
templates/->templates/AGENTS.md - editing
tests/or validation scripts ->tests/AGENTS.md
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.

