initiative-feasibility
Skill2 starsChanged 4 months agoMITView on GitHub
After the dev team has written their spec slice from the PRD, review it against the current codebase — baseline, gaps, impact, test harness, architecture governance (ADR + MDC), risks, and 4-lane triage. Writes the report locally and emits Forge readiness for the open Draft spec PR while Gate 2 remains spec-pending. Use when the dev has drafted the spec and wants to check buildability before technical review and planning.
In this repository22 other files it gives its agents.
- ground-specSkill
- learning-extractSkill
- loop-specSkill
- pre-implementSkill
- purge-initiative-artifacts-appSkill
- spec-draftSkill
- spec-implementation-planSkill
- spec-technical-reviewSkill
- ensure-repo-graphSkill
- post-product-questionsSkill
- prd-codebase-mapSkill
- review-product-questionsSkill
- commit-workspaceSkill
- create-board-ticketsSkill
- open-draft-prSkill
- prd-impact-mapSkill
- prd-qualitySkill
- prd-thinkSkill
- purge-initiative-artifacts-metaSkill
- review-findingsSkill
- update-documentsSkill
- validate-requirementsSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 9 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Never skip a check in references/checks.md.
- Distinctive: line 30
Don't fix — flag. Do not edit product source or product specs unless the user explicitly asks after the report.
- Distinctive: line 32
Run T0–T5 control loop (Gather → Understand → Analyze → Plan → Execute → Verify).
- Distinctive: line 33
Verify light source freshness before F1: product-spec header citations (H1 PRD digest, H2 scope digest, H3 map revision) and tip continuity.
- Distinctive: line 37
Do not require mid-lane report digests as staleness SSOT.
- Distinctive: line 38
Every open item must record lane, blocking, owner, status, required-by stage, default-if-deferred, evidence, and resolution reference.
- Very specific: line 47
If a codegraph provider is available (MCP tool matching prayog-skills/references/codegraph-tool-contract.md, or a local CLI), prefer it for …
- Distinctive: line 78
After dev updates spec — re-run on changed sections
- Distinctive: line 85
If the repo was removed, deferred, or its scope digest changed, stop and emit the map's ripple action.
- Very specific: line 112
Do not route PM/domain blockers to technical review via findings.
- Distinctive: line 124
Never carry a prior feasibility finding forward across changed H1–H3 authority without explicitly re-evaluating that finding.
- Very specific: line 143
Do not route engineering decisions to PM.
- Very specific: line 149
When the invocation binds handoff_path (orchestrator / AgentRunner baton), also overwrite that path with the same handoff: envelope before …
- Very specific: line 151
Follow prayog-skills/references/forge-side-effects.md#content-producers when this stage's pin has forge.commit_workspace other than …
- Very specific: line 158
Put lane counts, NEW-ADR, selected outcome rationale, and ripple action in signals.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Initiative feasibility
Assess whether the dev team's spec slice for this initiative is buildable in this repo as it exists today. Do not implement — flag only.
The spec slice is written by the dev team after reading the PRD. This skill reviews it for gaps, codebase alignment, and architecture governance.
Pattern borrowed from awesome-copilot create-github-issues-for-unmet-specification-requirements (extract → search code → classify). Evaluation discipline aligned with validate-requirements.
NON-NEGOTIABLE
- Never skip a check in references/checks.md. Mark SKIPPED with reason.
- Evidence for every finding — spec quote + repo path (file, symbol, or test name).
- Don't fix — flag. Do not edit product source or product specs unless the user explicitly asks after the report.
- Dual output: chat summary + saved report file + 4-lane triage.
- Run T0–T5 control loop (Gather → Understand → Analyze → Plan → Execute → Verify).
- Verify light source freshness before F1: product-spec header citations
(H1 PRD digest, H2 scope digest, H3 map revision) and tip
continuity; G1 meta head + approval while Gate 1 still applies. See
prayog-skills/references/artifact-write-contract.md. Stop on authority drift. Do not require mid-lane report digests as staleness SSOT. - Every open item must record lane, blocking, owner, status, required-by stage, default-if-deferred, evidence, and resolution reference.
- Read-only content skill. Persist the feasibility report locally and fill
handoff.forgefor/commit-workspace(or Gateflow ForgeClient) onto the same Draft spec PR branch opened by/spec-draft. Do not create branches, commit, push, open PRs, apply labels, create issues, merge, run probes, edit product source, or clean up environments. Gate 2 label staysspec-pending— do not setspec-lgtmduring feasibility. - Do not implement product code. Spec-only artifacts; publication is Forge.
- If a codegraph provider is available (MCP tool matching
prayog-skills/references/codegraph-tool-contract.md, or a local CLI), prefer it for architecture/impact/coverage questions. Always fall back to directsource_rootsreads when unavailable — never block or change outcome selection on its absence.
Inputs
Gather before starting. Resolve paths from .harness/profile.yaml or references/layout-defaults.md.
Dual workspace (orchestrated / Gateflow): workspace = app coding root
(spec slice + feasibility report). meta_workspace when bound = meta checkout
for PRD / impact-map freshness. Do not invent a meta path when empty.
- Initiative spec slice — primary doc (REQUIRED); written by dev team, lives under
product_spec_dirinworkspace - PRD — upstream requirements from
<client>-meta/prd/undermeta_workspacewhen bound (OPTIONAL; for conformance context; link from spec header) - Canonical impact map + approval evidence — (REQUIRED) exact revision and
tech-lead review referenced by the spec header (resolve under
meta_workspacewhen bound) - As-built —
implementation-status.md(REQUIRED) - Tests —
tests_readme,unit_tests_dir,live_verify_dir, and toolchain config from profile (REQUIRED) - Source — modules under
source_rootsfrom profile (REQUIRED) - Prior feasibility report — for incremental re-run (OPTIONAL)
- Layout —
.harness/profile.yamlor references/layout-defaults.md rules_glob— workspace MDC rules (REQUIRED). Read before T2 Analyze.adr_dir— architecture decision records (REQUIRED). Run relevant-ADR pass per references/governance.md before T2 Analyze.- Codegraph provider — OPTIONAL — see
prayog-skills/references/codegraph-tool-contract.md
When to use
- Dev has written spec slice and wants to check buildability before spec PR merge
- After dev updates spec — re-run on changed sections
- User asks: feasibility, impact, gap analysis, spec vs codebase
Process
- T0 Gather and freshness gate — inventory inputs; compare product-spec H1–H3 citations (and G1 when applicable) with live PRD/map/review. If the repo was removed, deferred, or its scope digest changed, stop and emit the map's ripple action. Feasibility report digests are walk-time only.
- T1 Understand — initiative id, spec branch, review objective
- T2 Analyze — read spec waves/capabilities; scan repo evidence; cross-reference
rules_globand relevant ADRs; flag spec wording that conflicts with MDC patterns or Accepted ADRs - T3 Plan — which checks run (full vs incremental)
- T4 Execute — run checks F1–F14 per references/checks.md
- T5 Verify — save report locally; publish summary + 4-lane triage; select workflow outcome from the lane-to-outcome rubric; fill Forge readiness
Outcome selection (lane → workflow)
Map evidence to exactly one outcome declared for initiative-feasibility in
pinned workflow.yaml. Prefer the first matching row:
| Outcome | When | Next (from workflow) |
stale | Product-spec H1–H3 / G1 / tip authority drift | spec-draft |
failed | Execution/analysis failure on otherwise valid inputs | workflow-stop |
blocked | Explicit approval/policy gate prevents progress (repo held, gate closed) | spec-human-decision |
needs-input | Unresolved blocking PM or domain item (missing product/SME answer) | spec-human-decision |
findings | Unresolved blocking PE / ADR / engineering item (Critical or Should-fix PE-lane) | spec-technical-review |
pass | Zero unresolved blocking findings; informational/Verify/Gap observations may remain | spec-technical-review |
Both pass and findings enter /spec-technical-review before plan (pin SSOT).
pass still means “no blocking PE findings at feasibility”; TDD may record
N/A / light confirmation. Do not skip technical review on clean feasibility.
Do not route PM/domain blockers to technical review via findings. Do not
emit pass while unresolved PE/ADR blockers remain. Informational observations
are report signals — they do not alone select findings.
Impact-map revision handling
continue— scope digest unchanged; record the newer revision check.re-draft/re-feasibility— scope digest changed; update the spec before running incremental feasibility.hold/close— repo removed, deferred, blocked, or approval stale.re-plan— dependency/build order changed; invalidate any existing plan.
Never carry a prior feasibility finding forward across changed H1–H3 authority without explicitly re-evaluating that finding. Feasibility report files are PURGE at initiative closure.
Output
Save to {reports_dir}/{feasibility_prefix}-{initiative}.md (from profile).
Use references/output-template.md.
Open items — 4-lane triage
End with four lanes (see output template):
- PM questions — product scope, UX, priority; comment on meta PRD PR (plain English)
- PE questions — engineering decisions, ADR gaps, test policy; comment on spec PR; resolved by
/spec-technical-review - Domain clarifications — business source-of-truth; route to named SME (meta PRD PR or issue)
- Auto-fixable — naming drift, inferred cross-references; record as findings/signals for later agent fix — do not mutate product source inside this skill
Do not route engineering decisions to PM. Apply the routing rubric in
the spec-technical-review skill's references/governance.md.
Workflow handoff
- Append/emit the envelope from
prayog-skills/references/handoff-envelope.mdto the saved report. Use stageinitiative-feasibility. - When the invocation binds
handoff_path(orchestrator / AgentRunner baton), also overwrite that path with the samehandoff:envelope before exit. Leaving the baton empty is a failed stage for automated consumers.artifact.pathremains the workspace skill output, not the baton path. Seeprayog-skills/references/handoff-envelope.md(Orchestrator baton). - Derive
next_candidatesandhuman_checkpointfrom pinned rootworkflow.yamlfor(stage: initiative-feasibility, outcome)perprayog-skills/references/handoff-envelope.md(Derive from pinned workflow). Sethuman_checkpoint: trueonly when the resolved next node'stypeishuman-checkpoint— never because the artifact "should be reviewed." - Follow
prayog-skills/references/forge-side-effects.md#content-producerswhen this stage's pin hasforge.commit_workspaceother thandisabledor next is anexternal-actionwithforge.requires— fillhandoff.forge/ recommend/commit-workspace; do not treat local CLI as skill success. Content skill success ≠ report published.
Transitions: pinned root workflow.yaml for this stage (SSOT) — including
pass vs findings routes. Human or agent may run this skill; legality and
auto-dispatch follow dispatch + delivery contract + latest handoff.
Put lane counts, NEW-ADR, selected outcome rationale, and ripple action in
signals; finding ids belong in blockers as FF-* (never bare F-12).
next_candidates never authorize invoke.
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.

