Consult pipeline
Skill Timelnapp/skills/skills/consulting-os/consult-pipeline
6 SKILLs to improve Claude Code memory, derived from Timeln's second brain capability to remember everything.
npx -y skills add Timelnapp/skills --skill consult-pipelineAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Trigger on "consult pipeline", "cos pipeline", "new prospect workflow", "run engagement pipeline", "build proposal end to end", "resume engagement", "/cos-plan", "/cos-resume". Orchestrates CAPTURE through SUMMARY with mandatory human checkpoints. Invokes existing consult-* and timeln-* skills at each stage — never duplicates Timeln logic. NOT a substitute for single-stage skills when user only wants one artifact.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
11.0 KB, ~2.9k tokens by cl100k_base, as published. Nobody here has run it
Consult Pipeline -- Solo-Founder Engagement Orchestrator
Runs the full pursuit workflow: CAPTURE → RESEARCH → FRAME → DESIGN → INTEGRITY → ARCHITECT → BUILD → REVIEW → PACKAGE → PURSUE → DELIVER → CLOSE → SUMMARY.
Human decides at every checkpoint. Max 2 revision loops between REVIEW and DESIGN. Reads/writes engagement-passport.yaml in the prospect folder.
Required skills (invoke — do not reimplement)
| Layer | Skills |
|---|---|
| Consult build | consult-frame, consult-arc, consult-gates, consult-acceptance, consult-commercial |
| Consult quality | consult-integrity, consult-consistency-lint, consult-red-team |
| Consult ship | consult-package, consult-pursue |
| Timeln (every stage — see table below) | timeln-find, timeln-decided, timeln-shipped, timeln-warned, timeln-quickly, timeln-plan |
| Architecture | arch-diagrams (external) |
Prospect folder layout
Cold-start pursue (/cos-pursue with client + topic only): all files flat under prospects/{slug}/ — see ../consult-pursue/cold-start/EXAMPLES.md.
Full pipeline (/cos-plan): nested under Output/{Client}-Engagement/:
prospects/{slug}/
├── engagement-passport.yaml
├── Input/ # transcripts, RFPs, briefs
└── Output/
└── {Client}-Engagement/
├── 00-research.md
├── 01-frame.md
├── 02-arc.md
├── 03-gates.md
├── 04-acceptance.md
├── 05-commercial.md
├── integrity-3.5.md
├── integrity-5.5.md
├── lint-report.md
├── red-team-report.md
├── {Client}-Engagement-Proposal-Pack.md
├── pursue/ # email, call script
└── Framework-Variants/
Copy skeleton from prospect-template/.
Engagement passport
Read at stage start; update stage, artifacts, checkpoints, revision_loop, timeln_pulls at stage end.
Env: COS_PASSPORT_RESET=1 — ignore existing passport; start at CAPTURE.
Resume: read stage field; run from that stage forward unless user says restart.
Timeln at every stage
Run the listed Timeln skill(s) before the consult skill for that stage. If MCP unavailable, note timeln skipped — no MCP in stage output and continue with source docs only — never fabricate Timeln results.
| Stage | Timeln skill(s) | Query / purpose |
|---|---|---|
| 0 CAPTURE | timeln-plan | Optional: if user asks "what pursuits this week?", run cascade first; else skip |
| 1 RESEARCH | timeln-find | Open synthesis: domain, client industry, analogous engagements |
timeln-decided | Settled calls on stack, delivery model, pricing patterns for this domain | |
| 2 FRAME | timeln-find | Stakeholder / industry context not in transcript |
timeln-quickly | Verbatim client quotes on decision, scope, constraints (Fact + Quote modes) | |
| 3 DESIGN — arc | timeln-shipped | Proof points for target capability and options |
timeln-warned | Past failures on proposed technical path | |
| 3 DESIGN — gates | timeln-decided | Prior phasing / gate patterns for similar engagements |
| 3 DESIGN — acceptance | timeln-warned | Fold real past failures into fallback rows (per consult-acceptance) |
| 3 DESIGN — commercial | timeln-decided | Precedent duration, access assumptions |
timeln-shipped | Precedent team shape / delivery evidence | |
| 3.5 INTEGRITY | timeln-quickly | Verify quoted client language against Timeln + source transcript |
timeln-shipped | Verify any "we've done X" claims in arc/commercial | |
| 4 ARCHITECT | timeln-warned | Gut-check architecture choices against past incidents |
timeln-decided | Prior ADRs on platforms, patterns named in diagram | |
| 5 BUILD | timeln-shipped | Credibility bullets for framework variants |
timeln-find | Analogous proposal structures / win themes from memory | |
| 5.5 REVIEW | timeln-warned | Feed real scars into consult-red-team Devil's Advocate |
| 6 PACKAGE | timeln-quickly | Optional: verbatim quotes for Q&A appendix when user asks |
| 7 PURSUE | timeln-shipped | Proof points for pitch email and call script |
timeln-quickly | Numbers, dates, names client mentioned | |
| 8 DELIVER | timeln-quickly | Mid-engagement recall (user invokes ad hoc) |
timeln-decided | Recall settled engagement decisions without relitigating | |
| 9 CLOSE | timeln-find | Capture learnings; suggest ingest titles for Timeln |
| 10 SUMMARY | timeln-plan | Optional: feed closed pursuit into next week's cascade |
Weekly (parallel, not a pipeline stage): cron timeln-plan — ranks which prospects get pipeline time.
Stage workflow
0 — CAPTURE
- Create
prospects/{slug}/from template if missing. - Initialize
engagement-passport.yaml(stage: CAPTURE). - Place input in
Input/(transcript, RFP, brief). - Checkpoint: "Is this a real pursuit worth framing?" → user yes/no.
1 — RESEARCH
timeln-find— domain + analog synthesis → section in00-research.md.timeln-decided— prior decisions on domain → same file under## Prior decisions.- Write
00-research.md; update passporttimeln_pulls. - Checkpoint: "Enough context to frame?"
2 — FRAME
timeln-find— stakeholder/industry context.timeln-quickly— client quotes on decision and scope.- Run
consult-frameusing Input + research + Timeln sections →01-frame.md. - Checkpoint: User approves frame card → set
checkpoints.frame_approved.
3 — DESIGN
Run in order; each step reads prior artifacts + relevant Timeln output:
- Arc —
timeln-shipped,timeln-warned→consult-arc→02-arc.md - Gates —
timeln-decided→consult-gates→03-gates.md - Acceptance —
timeln-warned→consult-acceptance→04-acceptance.md - Commercial —
timeln-decided,timeln-shipped→consult-commercial→05-commercial.md - Checkpoint: User picks option (MVP / Pilot / Full / custom) →
option_selected.
3.5 — INTEGRITY (pre-architecture)
timeln-quickly+timeln-shipped— prefetch for gate.- Run
consult-integritygate3.5→integrity-3.5.md. - BLOCK if verdict FAIL — return to DESIGN.
- Checkpoint: User acknowledges open items.
4 — ARCHITECT
timeln-warned,timeln-decidedon components in scope.- Run
arch-diagrams→ layered, sequence, deployment artifacts in Output folder. - Checkpoint: "Diagram matches approved scope?"
5 — BUILD
timeln-shipped,timeln-findfor variant credibility.- Generate framework variants per
../consult-arc/FRAMEWORK-VARIANTS.md. - Checkpoint: User picks variant or merge map → record in passport.
5.5 — REVIEW
timeln-warned— scars for red team.consult-consistency-lint→lint-report.mdconsult-red-team→red-team-report.md- If critical findings or red-team BLOCK: increment
revision_loop(max 2) → back to DESIGN or BUILD. - Checkpoint: Ship pack? yes/no
6 — PACKAGE
- Optional
timeln-quicklyfor appendix quotes. consult-package→{Client}-Engagement-Proposal-Pack.md- Checkpoint: Final ship approval →
checkpoints.ship_approved
7 — PURSUE
Standard (pack exists):
timeln-shipped,timeln-quicklyconsult-pursue→pursue/email.md,pursue/call-script.md- Checkpoint: Send to client?
Cold-start (user gives client + topic only — no pack, no Input brief):
- Infer slug; create
prospects/{slug}/if missing. - Web research →
prospects/{slug}/00-research-public.md. - Synthesize pack at
prospects/{slug}/{Client}-Engagement-Proposal-Pack.mdfrom../consult-pursue/cold-start/— label(synthesized). - Passport:
stage: PURSUE,option_selected: POC,source_mode: cold_start. timeln-shipped,timeln-quickly→consult-pursue.- Checkpoint: Send to client? Upgrade to full pipeline when brief arrives (
/cos-research).
See /cos-pursue and ../consult-pursue/cold-start/EXAMPLES.md.
Showcases: ../consult-pursue/examples/ho-brothers/ · ../consult-pursue/examples/vodafone-portugal/ (single-folder cold-start).
8 — DELIVER
Live engagement — not automated. User invokes timeln-quickly / timeln-decided ad hoc during delivery. Track gate sign-offs against 03-gates.md / 04-acceptance.md.
9 — CLOSE
timeln-find— synthesize learnings.- Write
09-close.md; suggest Timeln ingest (transcript, decision, retro). - Update passport
stage: CLOSE.
10 — SUMMARY
Write 10-process-summary.md: stages completed, revision loops, Timeln pulls used, time estimate, open items.
Optional timeln-plan to reprioritize next pursuits.
Post-win expansion (not a pipeline stage): once a deal is won, run /cos-market (consult-market) to find the next 5 lookalike accounts in the same industry + size for the same use case. Standalone, manual-seed — does not read the passport.
Output — process summary shape
## Pipeline summary -- {client} -- {YYYY-MM-DD}
**Stage reached:** ...
**Revision loops:** n/2
**Option selected:** ...
**Timeln skills invoked**
| Stage | Skill | Outcome |
|---|---|---|
| 1 | timeln-find | ... |
| ... | ... | ... |
**Artifacts**
- ...
**Open items**
- ...
**Next command:** /cos-resume | /cos-pursue | timeln-plan
Rules
- Never duplicate Timeln MCP calls inline — always invoke the timeln-* skill workflow.
- If a Timeln skill returns
no record, pass that through; do not invent memory. - Do not skip integrity 3.5 before ARCHITECT or 5.5 before PACKAGE.
- One primary decision thread from frame through pack — lint enforces.
- Max 2 revision loops; then escalate to user with explicit tradeoff list.
Commands
/cos-plan starts at 0; /cos-resume reads passport. (Slash commands in .claude-plugin/commands/.)
Pairing
| Skill | Pipeline role |
|---|---|
| timeln-plan | Weekly prioritization + optional post-close |
| timeln-find | Research, build, close synthesis |
| timeln-decided | Research, gates, commercial, architect, deliver |
| timeln-shipped | Arc, commercial, integrity, build, pursue |
| timeln-warned | Arc, acceptance, architect, review |
| timeln-quickly | Frame, integrity, package, pursue, deliver |
| consult-market | Post-win: top-5 lookalike target list (standalone) |
Gives 0 of the 12 instructions most plan spec skills give in ~2.9k tokens
Counted across 1,100 of the 1,860 authors here whose files we hold, read 2026-08-06
- ask one question at a timein 46 of 1100, across 38 files
- Break plans into vertical slicesin 28 of 1100, across 10 files
- Publish issues in dependency orderin 27 of 1100, across 9 files
- Iterate until user approves the breakdownin 24 of 1100, across 6 files
- Explore the repository to understand the codebase statein 24 of 1100, across 7 files
- Use domain glossary vocabularyin 23 of 1100, across 5 files
- Apply correct triage labels to published issuesin 23 of 1100, across 5 files
- Write failing tests before implementation codein 23 of 1100, across 18 files
- Prefer AFK slices over HITLin 22 of 1100, across 7 files
- ask clarifying questions until requirements are concretein 21 of 1100, across 13 files
- Respect existing architecture decision recordsin 20 of 1100, across 5 files
- write a specification before writing any codein 20 of 1100, across 12 files
Said here and by no other author read
- read the engagement passport at stage start
- update the engagement passport at stage end
- run the listed timeln skill before each consult skill
- stop for a human decision at every checkpoint
- allow a maximum of two revision loops
- copy the folder skeleton from the prospect template
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.