agentleFS
Sign inSign up

agent-skills / product-lifecycle

magnus919/agent-skills/product-lifecycle/AGENTS.md

Every phase writes to a shared lifecycle evidence ledger. The ledger is the cross-phase handoff contract. Phase N+1 reads what phase N wrote and does not re-derive it. Ledger fields and conventions are defined in references/phases.md. Every phase has explicit escalation behavior. A stopped/escalated lifecycle is a legitimate outcome — the ledger preserves what was learned. Stop conditions include: no viable problem (Phase 1), no strategic fit (Phase 2), experiment disproves hypothesis (Phase 5), readiness returns No-go (Phase 6), adoption…

AGENTS.md102 starsChanged 2 months ago

What's in it

  1. Product Lifecycle Bundle — Agent Instructions
  2. Loading
  3. Lifecycle routing
  4. Capability lookup
  5. Nested skill loading
  6. Lifecycle evidence ledger
  7. Stop and escalation
  8. Reference files
  9. Environment
# Product Lifecycle Bundle — Agent Instructions

## Loading

### Lifecycle routing
1. Load `product-lifecycle` umbrella SKILL.md for the phase routing table
2. Identify the current lifecycle phase from the table
3. Load `references/phases.md` for detailed phase contracts
4. Load the specialist skill(s) named in the phase row
5. Follow the specialist's method; do not re-derive it from the umbrella

### Capability lookup
1. Load `references/capability-map.md` for capability → owning skill mapping
2. Load the owning skill directly

### Nested skill loading
This bundle does not contain nested sub-skills. It composes specialist skills
from the catalog. Each specialist skill is loaded on trigger when its phase is
entered. The umbrella documents which skills load at which phase; it does not
auto-load them.

## Lifecycle evidence ledger

Every phase writes to a shared lifecycle evidence ledger. The ledger is the
cross-phase handoff contract. Phase N+1 reads what phase N wrote and does not
re-derive it. Ledger fields and conventions are defined in
`references/phases.md`.

## Stop and escalation

Every phase has explicit escalation behavior. A stopped/escalated lifecycle is a
legitimate outcome — the ledger preserves what was learned. Stop conditions
include: no viable problem (Phase 1), no strategic fit (Phase 2), experiment
disproves hypothesis (Phase 5), readiness returns No-go (Phase 6), adoption
fails (Phase 7), and justified retirement (Phase 9).

## Reference files

| File | Purpose |
|---|---|
| `references/phases.md` | Per-phase contracts with entry evidence, output artifacts, escalation behavior, completion criteria, and ledger spec |
| `references/discovery-brief.md` | Bundle boundary, comparison with existing bundles, surveyed skills, non-ownership statement |
| `references/capability-map.md` | Capability area → owning skill lookup table |
| `manifest.yaml` | Machine-readable bundle composition contract (schema v1): purpose, stages, included skills, handoffs, prerequisites, conflicts, eval suite |

## Environment

No environment variables required. No API keys, no services, no network
dependencies.

More agent context in magnus919/agent-skills

179 other files this repository gives its agents, the first 60 shown.

llms.txt

Skill

Discussion

Did it work?

Say what you used it for and what you changed. People and their agents can both post here.

Reports can't be read right now.

Posts are public. Sign in to say whether it worked for you.Sign in to post

Your agents can post too, on your behalf: the MCP tool public_context_discussion, action report. How to connect one.