Agent Harness¶

Spec-driven development (SDD) frameworks and agent harnesses — the layer above agent skills that owns the requirements → spec → plan → tasks → execute → verify loop, manages context windows, or runs sub-agents.

This repo focuses on skills and intentionally does not ship a SDD framework or harness. It does ship a local skill for designing/reviewing the LLM application layer, plus Herdr's official control adapter: neither turns this collection into a harness runtime.

Skills in this repo¶

Local¶

Skill	One-line	Notes
`12-factor-agent-design-review`	Design or evidence-review production LLM applications using HumanLayer's 12-Factor Agents principles	Reviews the workflow/application layer below a harness: prompts, context, typed decisions, state, control flow, pause/resume, humans, retries, and replay. It does not provide a runtime.

Vendored¶

Skill	Upstream	Series
`herdr`	`herdrdev/herdr`	flat

External skills (manual install)¶

Status enum (single source of truth across the catalog)

Every external-skill or MCP entry below carries a status: value:

Status	Meaning	Where it links
`vendored`	Already in `vendor.yaml`.	Link to the `skills/vendor/<name>/` (or series) entry.
`deferred`	Open `TODO P?` item — under consideration.	Link to the `TODO.md` anchor.
`skipped`	Looked at, chose not to vendor.	Inline reason required (1 sentence).
`evaluated`	Read but no decision recorded. Effectively "tracked."	Inline 1-line note describing what was learned.
`wishlist`	Surfaced but not yet evaluated.	No link required; default for fresh discoveries.

Status changes are explicit edits — see the Adding catalog entries workflow doc (under Workflows in the nav) for the wishlist → deferred → vendored recipe.

Manual install for non-vendored skills

When a skill is not in vendor.yaml, you can still install it directly from upstream via the npx skills CLI:

# Whole-repo install (if SKILL.md sits at repo root or skills/)
npx skills@latest add <owner>/<repo>

# Subpath install (when the skill is nested)
npx skills@latest add <owner>/<repo>/<path-to-skill-dir>

# Cherry-pick by name
npx skills@latest add <owner>/<repo> -s <skill-name>

# Force deep discovery (skill nested >1 level inside skills/)
npx skills@latest add <owner>/<repo> --full-depth -s <skill-name>

For Anthropic-style plugin marketplaces (e.g. anthropics/financial-services, anthropics/knowledge-work-plugins), use the Claude Code plugin commands instead:

claude plugin marketplace add <owner>/<repo>
claude plugin install <plugin-name>@<repo-name>

For MCP servers, see the per-entry "Auth & install" section in the MCP wiki — install commands vary by transport (HTTP / stdio / SSE) and host (Claude Code / Claude Desktop / Cursor / Managed Agents).

Skill / Plugin	Upstream	Status	Why this status	Install hint
`spec-kit` (skills mode)	`github/spec-kit`	`evaluated`	The de facto SDD framework (95.5k ⭐). 30+ supported agents. Has a `--skills` install mode that ships its slash commands as agent skills. Documented in `reference/sdd-and-harnesses.md`. Not vendored — would compete with the per-skill philosophy of this repo.	`uvx --from git+https://github.com/github/spec-kit specify init <project> --integration claude-code --integration-options="--skills"`
`get-shit-done` (gsd)	`gsd-build/get-shit-done`	`evaluated`	Earlier, lighter SDD (61.4k ⭐). 6-command loop (`gsd-new-project → gsd-discuss-phase → gsd-plan-phase → gsd-execute-phase → gsd-verify-work → gsd-ship`).	(see upstream README)
`gsd-2`	`gsd-build/gsd-2`	`evaluated`	Standalone harness (not just a SDD framework).	(see upstream README)
OpenClaw	`openclaw/openclaw`	`evaluated`	Standalone CLI/runtime that controls the agent session. Also produces `gstack-openclaw-*` skills (already vendored under `product-planning` series).	(see upstream README)
Pi SDK	`badlogic/pi-mono`	`evaluated`	Agent harness alternative.	(see upstream README)
`agent-architecture-analysis`	`existential-birds/beagle`	`skipped`	Evidence-first 12-Factor review is useful, but the rubric is Python- and implementation-specific; the local skill cites only the portable evidence-gate idea.	`npx skills add existential-birds/beagle --skill agent-architecture-analysis`
12-factor agent skill pack	`tika/12-factor-agent-skills`	`skipped`	Broad five-skill pack, but cross-skill references and scanner self-matches make it unsuitable for vendoring as-is.	`npx skills add tika/12-factor-agent-skills`
pstack (`poteto-mode`)	`cursor/plugins/pstack`	`evaluated`	Opinionated Cursor workflow pack with a router, playbooks, principle skills, and subagent fan-out. Too coupled and fast-moving to vendor. Claude Code is supported upstream through a generated adapter. For Codex and others, use community ports (`Go7hic/ystack`, `TheOnlyFusionCube/potetos-for-everyone`). See skill collections.	`/plugin marketplace add cursor/plugins` then `/plugin install pstack@cursor-plugins`

MCP servers¶

Name	Upstream	Status	Auth	Records
not applicable — harnesses use whatever MCPs the underlying agent has access to

Backlog (TODO `P?` items)¶

Harness runtimes remain out of scope; only their reusable skill adapters are eligible for vendoring.