feat(skill): rename to agentfield, fetch live docs from agentfield.ai#611
Merged
Merged
Conversation
…m agentfield.ai
Renames the bundled `agentfield-multi-reasoner-builder` skill to plain
`agentfield` and rewrites it around three changes:
- **Live docs first**: every invocation now fetches
agentfield.ai/llms.txt + llms-full.txt + per-page /llm/docs/<slug>
before scaffolding, so the agent gets ground-truth SDK signatures
instead of a baked-in snapshot that drifts release to release. The
bundled references/primitives-snapshot.md is the offline fallback,
explicitly stamped as such.
- **Pattern bias removed**: architecture-patterns.md becomes
patterns-emerge.md. Patterns are vocabulary for naming what falls out
of the five principles, not a menu to copy from. HUNT->PROVE has
equal weight with linear refinement, dynamic router, fan-out/recurse,
event-driven, etc. The new examples-map.md indexes
code/examples/* so the agent picks the closest real build and greps
its actual code instead of paraphrased templates.
- **CLI introspection surface**: new references/cli-toolkit.md
documents `af doctor`, `af init`, `af agent {status,discover,query,
run,agent-summary,kb,batch,help}`, `af logs`, `af execution`, etc.
The skill now prefers `af agent kb guide --goal "..."` and
`af agent discover -q "..."` over hand-rolled curl.
Other changes:
- New references/model-selection.md: env -> ask user -> live-pick from
openrouter.ai/api/v1/models (open-weight providers) -> frozen
fallback. No more hardcoded model string in nine places.
- SKILL.md adds a dedicated "Reasoners are APIs — design like a service
mesh, not a chain" section explaining what no static chain framework
(LangChain, CrewAI, AutoGen, LangGraph) can do.
- scaffold-recipe.md depersonalized: no committee.py / adversarial
reviewer / HUNT->PROVE example. Hard invariants only.
- anti-patterns.md is now a one-line table.
- All bundled refs trimmed.
Compat:
- Catalog adds an `Aliases` field; "agentfield-multi-reasoner-builder"
still resolves to the new skill so existing installs keep working.
- Version bumped 0.3.0 -> 0.4.0.
Verified:
- go build ./control-plane/... + go vet ./...
- go test ./internal/skillkit/... ./internal/cli/... (384 pass)
- ./scripts/sync-embedded-skills.sh --check (no drift)
- `af skill catalog` shows agentfield v0.4.0
- `af skill print agentfield-multi-reasoner-builder` resolves via alias
Contributor
📊 Coverage gateThresholds from
✅ Gate passedNo surface regressed past the allowed threshold and the aggregate stayed above the floor. |
Contributor
📐 Patch coverage gateThreshold: 80% on lines this PR touches vs
✅ Patch gate passedEvery surface whose lines were touched by this PR has patch coverage at or above the threshold. |
… name Four cosmetic updates to scripts/install.sh that referenced the legacy agentfield-multi-reasoner-builder name. The install logic itself is unchanged — it always installed whatever the af binary's catalog defaults to.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Renames the bundled skill from
agentfield-multi-reasoner-builderto plainagentfieldand rewrites it around three goals:agentfield.ai/llms.txt,llms-full.txt, and per-page/llm/docs/<slug>before scaffolding. The bundled snapshot (primitives-snapshot.md) is the offline fallback only, explicitly stamped as such — so the agent stops teaching coding agents stale SDK signatures when the SDK ships a release.architecture-patterns.md→patterns-emerge.md. Named patterns (HUNT→PROVE, parallel hunters, streaming, meta-prompting, factory loops, linear refinement cascade, dynamic router, reactive enrichment, etc.) are now vocabulary for naming what falls out of the five principles. HUNT→PROVE no longer dominates the prose. A newexamples-map.mdindexes every project incode/examples/*with one-line shape descriptions so the agent picks the closest live build and greps its real code instead of paraphrased templates.cli-toolkit.mddocumentsaf doctor,af init,af agent {status,discover,query,run,agent-summary,kb,batch,help},af logs,af execution,af vc. The skill now prefersaf agent kb guide --goal "..."andaf agent discover -q "..."over hand-rolled curl.Other changes
model-selection.md:af doctor→ ask user → live-pick fromopenrouter.ai/api/v1/models(filtered for open-weight providers) → frozen fallback. No more hardcodedgemini-2.5-flashin nine places.SKILL.mdadds a dedicated "Reasoners are APIs — design like a service mesh, not a chain" section that explicitly contrasts with LangChain / CrewAI / AutoGen / LangGraph static DAGs.scaffold-recipe.mddepersonalized: removedcommittee.py/ adversarial reviewer / HUNT→PROVE inline example. Hard invariants only.anti-patterns.mdcompressed to one-line tables.Back-compat
Skillstruct gets a newAliases []stringfield.agentfield-multi-reasoner-builderresolves transparently to the new skill, so anyone with the old skill installed keeps working.0.3.0→0.4.0.File-size comparison
SKILL.mdVerified
go build ./control-plane/...go vet ./...(no issues)go test ./internal/skillkit/... ./internal/cli/...(384 passed)./scripts/sync-embedded-skills.sh --check(no drift)af skill catalogshowsagentfield v0.4.0af skill print agentfield-multi-reasoner-builderresolves via aliasTest plan
SKILL.mdand confirms tone + lengthlive-docs.md+cli-toolkit.md(the two largest new docs)af skill print agentfield-multi-reasoner-builder)go build,go test, sync-embedded-skills--check🤖 Generated with Claude Code