agentsclimarketplace

Qualifying opportunities

Skill danielkinneyspears/govcon-pursuit-brain/skills/qualifying-opportunities

Wiki-native Claude Skills for the US Federal capture and proposal lifecycle. Karpathy's LLM Wiki pattern applied to GovCon pursuits, with a comprehensive federal-acquisition domain wiki. Apache-2.0.

Install
npx -y skills add danielkinneyspears/govcon-pursuit-brain --skill qualifying-opportunities

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 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

Runs a structured bid/no-bid gate review for a US Federal opportunity and bootstraps the pursuit's knowledge wiki. Use when deciding whether to pursue or bid a federal solicitation, RFP, RFQ, RFI, sources-sought notice, or IDIQ task order — for a bid/no-bid decision, pursuit decision, gate review, opportunity qualification, go/no-go, or pWin assessment. Creates the pursuit wiki and its schema, builds the customer and bid-decision entities, and renders the opportunity-profile and bid-decision views.

SKILL.md

10.2 KB, as published. Nobody here has run it

Qualifying Opportunities

Decide whether to pursue or bid a US Federal opportunity — and, because this is usually the first skill to touch a pursuit, bootstrap the pursuit wiki it will live in. The expensive mistake in federal BD is spending bid-and-proposal money on bids that were never winnable; this skill makes that decision deliberately, on evidence, and records it as the first entities in the pursuit's knowledge graph.

When to use this skill

Use this skill at any bid/no-bid gate — an early pursuit decision, a preliminary bid decision, or post-RFP bid validation. It both renders the decision and, on first run, creates the pursuit wiki that every later skill builds on.

It does not build the win plan — that is planning-capture and developing-win-strategy. It decides whether there should be a win plan at all.

Architecture

This skill is wiki-native. Read docs/architecture.md and the conventions before running it. The skill reads sources and the domain wiki, maintains pursuit entity pages, and renders views — it never writes a flat artifact as a source of truth.

Relevant domain-wiki pages: [[uniform-contract-format]], [[section-m]], [[far-part-15-source-selection]], [[best-value-tradeoff]], [[lpta]].

Inputs

Required: the opportunity — a described lead, or a solicitation document placed in pursuits/<id>/sources/.

Existing pursuit wiki, if this is not the first run: read pursuit.md and the existing customer/, competitors/, and decisions/ entities; do not re-create them.

Private company context (recommended). If pursuit.md carries a company_context_path: field, read the relevant company-context pages before intakecustomers/<this-agency>/, past-performance/, key-personnel/, pricing-guidance/, compliance-posture/, debrief-lessons/, approved-proof-points/. Treat them as authoritative sources for the questions they answer. See ../../docs/customize-for-your-company.md. Without company-context, the skill still runs from intake — just with less leverage.

Intake

Ask these as a numbered list. Skip questions whose answers are already in the solicitation, an existing pursuit entity, or the configured company-context. When company-context answers a question, cite the company-context page on the resulting entity (use its path as source:, inherit its confidence, tag the resulting entity sensitivity: company-proprietary). Unknowns are valid answers and lower confidence; never guess.

  1. The opportunity. Agency and program office, scope, contract type, estimated value, period of performance, set-aside, solicitation number. Is a document available?
  2. The gate and posture. Which gate is this (pursuit decision / bid decision / bid validation)? What is the team's competitive posture — incumbent, challenger, teaming partner, outsider?
  3. Customer relationship. Prior relationship with this customer and office; any shaping or engagement to date; the incumbent, and whether it is the team.
  4. Capability and past performance. Can the team perform the work? Is there recent, relevant past performance? Are required clearances, certifications, and the NAICS size standard met?
  5. Competition and funding. Likely bidders; realistic competitiveness; is funding confirmed; known or estimated budget?
  6. Strategic fit and blockers. Strategic alignment; blockers — OCI, a teaming dependency, a missing vehicle, capacity, unacceptable terms.

Workflow

Bid/No-Bid Gate Review:
- [ ] Step 1: Bootstrap or open the pursuit wiki; read company-context if configured
- [ ] Step 2: Complete intake; record unknowns as gaps
- [ ] Step 3: Build the customer entities
- [ ] Step 4: Run knockout checks and the weighted scorecard
- [ ] Step 5: Estimate pWin; form the recommendation
- [ ] Step 6: Write the bid-decision entity + supporting claims/ and source-register/ entities
- [ ] Step 7: Render the views and validate

Step 1: Bootstrap or open the pursuit wiki; read company-context. If pursuits/<id>/pursuit.md does not exist, perform the workspace-safety check (confirm the location is not in tracked version control; see the pursuit .gitignore), then create the pursuit directory and pursuit.md from schema/pursuit-schema.template.md — setting identification, posture, mode (full or lite), the sensitivity policy, and the company_context_path: if a private company-context workspace is configured. If it exists, read it. If a company_context_path is set, read the relevant company-context pages now (customers, past performance, key personnel, pricing guidance, compliance posture, debrief lessons, approved proof points) — they will short-circuit several intake questions.

Step 2: Intake. Work the intake questions. Carry every unknown forward as a gap; an honest unknown beats a confident guess.

Step 3: Customer entities. Create entities/customer/ pages — the agency, the program office, and one customer/stakeholders/<name>.md per known decision-maker or influencer. Tag each with source, confidence, sensitivity, provenance. In lite mode, the customer may be a single page.

Step 4: Knockouts and scorecard. Run the knockout checks and the weighted six-category scorecard in references/bid-no-bid-criteria.md. A triggered knockout drives the recommendation regardless of score. Every score carries one line of evidence; where the evidence is an unknown, score conservatively.

Step 5: pWin and recommendation. Estimate pWin per the criteria file, applying its anti-optimism discipline. Form one recommendation: bid, no-bid, or conditional bid (with specific, assigned, dated conditions).

Step 6: Bid-decision entity + claims and source-register. Write entities/decisions/bid-decision.md: the gate, the knockout results, the scorecard, pWin, the recommendation, the rationale, and the gaps. Link it to the customer entities and to any competitors/ stubs. Tag it approval: pending — a bid decision is a high-risk output awaiting a human gate decision. If competitors were named, create low-confidence competitors/<name>.md stubs for analyzing-competitors to enrich later.

Then build the supporting evidence graph:

  • entities/claims/: write one claim entity per assertable fact the bid decision rests on. Examples: "Customer is the same buying office that bought the prior contract," "Team holds an active SBA 8(a) certification," "Funding line is confirmed in agency budget request." Each claim entity carries the claim text, the owner who can defend it, the assertion type (known fact / assessed / hypothesis), the proposal use, and a status.
  • entities/source-register/: write one source entity per external source the claims rest on. SAM.gov notices, USAspending records, agency budget documents, prior award notices, customer meeting notes (from the pursuit's contact reports), company-context pages. Each source entity carries the source URL or path, authority (which rung of the _source-policy.md ladder), directness, recency, independence, and a date pulled.

Every claim links to one or more sources; every source lists the claims it supports. Confidence on the bid-decision entity is then the lowest claim confidence it rests on.

Step 7: Render and validate. Render two views to pursuits/<id>/views/: opportunity-profile.md and bid-decision.md, per conventions/views.md. Run the validation checklist in references/bid-no-bid-criteria.md. Update pursuit.md's entity index, view index, and change log.

Output

A bootstrapped pursuit wiki: pursuit.md (with company_context_path if configured), the customer/ entities, the decisions/bid-decision.md entity, supporting claims/ and source-register/ entities, optional competitors/ stubs, and the opportunity-profile.md and bid-decision.md views.

Tell the user: the recommendation, the score and pWin, the top decision drivers, the most important gaps, and that the bid decision is approval: pending.

References

  • references/bid-no-bid-criteria.md: the knockout checks, the weighted six-category scorecard, pWin estimation, and the validation checklist.

Guardrails

  • Bootstrap safely. Before creating the pursuit wiki, confirm its location is not tracked in version control; a pursuit wiki holds sensitive material.
  • Honest unknowns beat confident guesses. Never invent a relationship, a competitor read, or funding certainty. Record gaps; set confidence honestly.
  • Knockouts are not negotiable. A real disqualifier yields a no-bid (or a conditional bid contingent on resolving it), regardless of score.
  • The bid decision is decision support. It is approval: pending until a human gate owner decides. The skill informs the call; it does not make it.
  • Maintain the graph, render the view. Entities are the source of truth; the views are rendered read-outs and are never hand-edited.
  • Company-context stays private. When the skill draws on a configured company_context_path, it references the company-context pages by path; it does not copy their content into the pursuit wiki. The resulting pursuit entities carry sensitivity: company-proprietary where the underlying fact came from company-context. See ../../docs/customize-for-your-company.md.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.