agentleFS
chakmarebelchakmarebel/federal-proposal-copilot/.claude/skills/import-from-capture/SKILL.md

import-from-capture

Skill4 starsChanged 3 months agoMITView on GitHub

Import a fully-qualified opportunity from the capture-pipeline and scaffold a proposal workspace pre-populated with solicitation facts, Go/No-Go assessment, customer POCs, and competitive intel. Use when a Prospects → Pursuing decision has been made in capture-pipeline and you want to start the proposal without re-typing the context.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After the user has clicked 📤 Export for Proposal on the opportunity detail page in capture-pipeline

    line 27
  2. Distinctive:

    Fetch from a deployed instance URL: if the user has a capture-pipeline URL + opportunity id, this skill can fetch the JSON directly via …

    line 37
  3. Distinctive:

    Ask the user (skip any that's already provided):

    line 55
  4. Distinctive:

    If validation fails: stop, report what's missing, don't proceed.

    line 78
  5. Very specific:

    Read my-company/capture-profile.md if it exists.

    line 96
  6. Very specific:

    Use the same scaffold contract as /new-proposal — refer to .claude/skills/new-proposal/SKILL.md (Steps 2, 3, 3b) for the authoritative …

    line 111
  7. Very specific:

    If no registry type clearly fits, exit the import with: "Unable to auto-determine proposal type from capture payload.

    line 167
  8. Distinctive:

    Write the following files from the capture payload.

    line 171
  9. Very specific:

    Write working/proposal-brief.md using the standard template (see /new-proposal), but pre-fill what the capture payload knows:

    line 249
  10. Distinctive:

    If the short name collides, suggest an alternative and confirm.

    line 319

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Import from Capture

Purpose

Bridge the handoff between the capture-pipeline (upstream: identify, qualify, decide to pursue) and the federal-proposal-copilot workspace (downstream: write the proposal). The capture side already knows the opportunity intimately — solicitation facts, NAICS, set-aside, scoring, Go/No-Go assessment, suggested teaming partners, customer POCs, next steps, and activity history. This skill pulls all of that across in one shot and pre-populates the proposal workspace so /proposal-manager has capture intel without re-deriving anything.

When to Use

  • A Prospects → Pursuing decision has been made in capture-pipeline
  • User says "import this opportunity", "pull in capture intel", "start proposal from capture", "import from BD pipeline"
  • After the user has clicked 📤 Export for Proposal on the opportunity detail page in capture-pipeline

Prerequisites

  • /setup-company has been run (my-company/ exists)
  • User has a capture export JSON file ready. Three ways to get one:
    1. Download from the web UI: click the 📤 Export for Proposal button on the opportunity detail page; browser saves a capture-<slug>-<id>.json file.
    2. Fetch from a deployed instance URL: if the user has a capture-pipeline URL + opportunity id, this skill can fetch the JSON directly via WebFetch (no manual download needed).
    3. Paste the JSON into the chat; this skill parses it inline.

Inputs

One of the three above. The skill accepts:

  • A local file path: ./downloads/capture-ai-edge-124.json, %USERPROFILE%\Downloads\capture-....json, etc.
  • A URL to the export endpoint: https://bd.example-pipeline.com/opportunity/124/export
  • A JSON payload pasted directly into the chat

Workflow

Step 1: Obtain the Capture Payload

Ask the user (skip any that's already provided):

  1. Which capture-pipeline instance and opportunity? (Optional — if they give you a URL or file path, proceed.)
  2. File path OR URL OR paste the JSON.

Resolve the input:

  • File path → use the Read tool. Confirm the file exists; if not, suggest they re-download from the detail page.
  • URL → use the WebFetch tool. URL must end in /opportunity/<id>/export. Tell the user that Cloudflare Access or app-password auth may block the fetch — if so, they need to download via the browser button instead.
  • Pasted JSON → parse directly from the chat.

Validate:

  • Top-level object has schema_version, opportunity, scores.
  • schema_version starts with 1. (major-version bump means the schema changed and this skill needs updating — warn the user).
  • opportunity.title is non-empty.

If validation fails: stop, report what's missing, don't proceed.

Step 2: Propose a Short Name

Derive a short directory name from opportunity.title using these rules:

  • Lowercase, replace non-alphanumeric with hyphens, collapse runs
  • Truncate to 40 chars
  • Drop leading/trailing hyphens

Show the user the proposed short name and confirm before creating directories. They can override (e.g., shorter, more memorable). This is the directory under proposals/ — irreversible without a rename, so get agreement.

Step 3: Confirm Capture Mode

Read my-company/capture-profile.md if it exists.

Map the capture-pipeline notice_type + scores.rules_status to a sensible default:

  • SBIR / CSO / OTA / BAA / White Paper (any notice type matching) → Responsive
  • Full & Open RFP / IDIQ Task Order / Recompete → Full Capture
  • Anything with rules_status == "sub" → Responsive (we're the sub, the prime drives the full capture effort)

Confirm with user before proceeding.

Step 4: Scaffold the Proposal Directory

Use the same scaffold contract as /new-proposal — refer to .claude/skills/new-proposal/SKILL.md (Steps 2, 3, 3b) for the authoritative layout and seeding logic. Re-stating here only for visibility; if the two ever drift, /new-proposal is canonical:

proposals/<short-name>/
├── inputs/
│   ├── 00_priority/
│   ├── 01_customer/
│   ├── 02_yourCompany/        ← seed from my-company/ (NOT 02_company)
│   ├── 03_teammates/
│   ├── 04_patterns/
│   ├── 05_graphic_standards/
│   └── 06_notes/
├── working/
├── drafts/
├── graphics/
├── reviews/
└── final/                     ← scaffolded for /export-proposal
    ├── docx/
    ├── xlsx/
    ├── pptx/
    ├── pdf/
    └── graphics-png/

Seed inputs/02_yourCompany/ the same way /new-proposal does:

  • Copy my-company/company-description.md
  • Copy my-company/capabilities.md
  • Copy my-company/contract-vehicles.md
  • Copy my-company/past-performance.md
  • Copy my-company/components-table.md (if present)

Copy my-company/distribution-statements.md → inputs/00_priority/ (falling back to reference/distribution-statements.md if the company-specific version is absent).

Step 4b: Seed proposal-type.md and activity.md (mandatory — current contract)

This step was missing in the original import-from-capture flow and is mandatory now. Without it, downstream skills (/status, /compliance-check, /red-team-review, /pricing-analyst, /export-proposal) cannot determine the proposal type and will refuse to run.

  1. Determine the proposal type from the capture payload. The capture payload's opportunity.notice_type (or parent_solicitation_number pattern) typically maps to a registry type:
    • Solicitation / Combined Synopsis/Solicitation with FAR clauses → far-rfp
    • Commercial Solutions Opening (CSO) → cso-brief (Phase 1) or cso-full (Phase 2)
    • Broad Agency Announcement → baa
    • Other Transaction notices → ota-white-paper or ota-proposal
    • SBIR/STTR Topic Notice → sbir-phase1 or sbir-phase2
    • Sources Sought → sources-sought
    • Request for Information → rfi
    • Request for White Papers (unsolicited context) → white-paper
    • Ambiguous → ask the user, list the registry from reference/proposal-types/
  2. Copy the chosen type file: reference/proposal-types/<type_id>.md → working/proposal-type.md. Do NOT edit the registry file. If the user overrides anything (e.g., "treat as Full Capture"), add an # Overrides section below the frontmatter of the copy only.
  3. Seed working/activity.md: copy templates/working/activity.md → working/activity.md, then append:
    ## <timestamp> — import-from-capture — imported opp <opportunity_number> from capture-pipeline, type=<type_id> → proposals/<short-name>/

If no registry type clearly fits, exit the import with: "Unable to auto-determine proposal type from capture payload. Please run /new-proposal manually and pick from the registry." Do not scaffold a workspace without proposal-type.md.

Step 5: Pre-Populate Capture Intel

Write the following files from the capture payload. Each one contains structured content the downstream skills can read as-is (no further parsing required).

inputs/00_priority/what-matters.md — drop in payload.proposal_kickoff_markdown unchanged. The capture-pipeline already rendered this as a clean markdown summary with opportunity basics, financials, Go/No-Go assessment, and next steps. /proposal-manager reads this first as its capture-intel anchor.

inputs/00_priority/solicitation-facts.md — build from payload.opportunity:

# Solicitation Facts

- **Title:** {opportunity.title}
- **Solicitation #:** `{opportunity.solicitation_number}`
- **Parent CSO/BAA:** `{opportunity.parent_solicitation_number}` (if different)
- **Agency:** {opportunity.agency}
- **Office:** {opportunity.office}
- **Notice type:** {opportunity.notice_type}
- **NAICS:** {classification.naics_code}
- **Set-aside:** {classification.set_aside}
- **Posted:** {opportunity.posted_date}
- **Response deadline:** {opportunity.response_deadline}
- **SAM.gov link:** {opportunity.url}

## Solicitation Summary (from SAM.gov)

{opportunity.summary}

## Description

{opportunity.description}

inputs/01_customer/customer-pocs.md — from payload.next_steps.customer_pocs:

# Customer POCs (from capture intel)

These were identified during the capture phase. The `/customer-intel`
skill will enrich this with agency mission, buying history, and
relationship gaps.

{for each POC:}
## {poc.name} — {poc.title}

- **Role in process:** {poc.role_in_process}
- **Source:** {poc.source}  _(solicitation = named in RFP; inferred = surfaced by AI)_
- **Notes:** {poc.notes}

Skip this file if next_steps.customer_pocs is empty.

inputs/03_teammates/suggested-partners.md — from payload.next_steps.partners:

# Suggested Teaming Partners (from capture intel)

AI-generated candidates from the capture-pipeline's quick-vet.
Review critically — these are hints, not commitments.

{for each partner:}
## {partner.name} — {partner.role} ({partner.strength} fit)

{partner.rationale}

Skip if empty.

working/capture-intel.json — save the full JSON payload unchanged. Provides a durable audit trail and lets any future skill re-parse the original capture context if it needs detail that didn't make it into the rendered markdown.

Step 6: Write the Proposal Brief

Write working/proposal-brief.md using the standard template (see /new-proposal), but pre-fill what the capture payload knows:

# {opportunity.title}

## Basics
- **Short name:** <short-name>
- **Proposal type:** <inferred — SBIR / CSO / OTA / BAA / etc. — from notice_type>
- **Customer:** {opportunity.agency} / {opportunity.office}
- **Due date:** {opportunity.response_deadline}
- **Created:** {today}
- **Source:** Imported from capture-pipeline (opp id {opportunity.id})

## Capture Mode
**<Responsive | Full Capture>**  ← confirmed in Step 3

## Capture Context
- **Rules score:** {scores.rules_score} ({scores.rules_status})
- **AI fit score:** {scores.llm_score}/10 — {scores.llm_action}
- **Prime feasibility:** {assessment.prime_feasibility.verdict}
- **Sub feasibility:** {assessment.sub_feasibility.verdict}
- **Technical alignment:** {assessment.technical_alignment.verdict}
- **Competitive position:** {assessment.competitive_position.verdict} (incumbent risk: {assessment.competitive_position.incumbent_risk})

## Financial Shape (AI-inferred, confirm before pricing)
- **Estimated TCV:** {financials.tcv_display}
- **Period of performance:** {financials.period_of_performance.value}
- **Contract vehicle:** {classification.contract_vehicle.value}
- **Incumbent:** {context.incumbent.value}

## Key Questions
1. What problem does the customer have?
2. What is our unique value for this customer?
3. What specific commitments can we make?
4. What proof points are most relevant?
5. Who is the decision maker?

## Next Steps
1. Review `inputs/00_priority/what-matters.md` — the capture Go/No-Go
   assessment and next-step recommendations are there.
2. Drop the full solicitation PDF into `inputs/00_priority/` (the
   capture payload has the SAM.gov summary but not the full text).
3. Run `/customer-intel` to enrich POCs with agency mission and
   buying history.
4. Run `/proposal-manager` to analyze the solicitation and build the
   compliance framework. It will read the capture intel automatically.
5. Continue the standard workflow (architect → past performance →
   pricing → graphics → writer → red team).

Step 7: Confirm and Report

Tell the user:

  1. What was created and where.
  2. What files were pre-populated vs. still empty.
  3. What's missing that they should add:
    • Full solicitation PDF (the capture summary is a truncated excerpt)
    • Any additional customer context they have from prior relationships that the LLM couldn't know
    • Teammate POCs or agreements
  4. Which skill to run next — usually /proposal-manager directly (capture intel is richer than a cold start, so we can skip /customer-intel on the first pass).

Rules

  • Never overwrite an existing proposals/<short-name>/ directory. If the short name collides, suggest an alternative and confirm.
  • Preserve the full capture JSON in working/capture-intel.json for downstream reference — don't discard data because a specific skill doesn't need it today.
  • Don't fabricate fields missing from the payload. If assessment is null, say so in the brief ("Capture assessment not available — run AI Score on the opportunity in capture-pipeline and re-export"); don't generate a fake one.
  • Respect AI-inferred source tags. Fields with source: ai_quickvet should be flagged as needing confirmation in the proposal brief and working files — they're starting points, not facts.

Output Format

At the end, emit a compact status summary:

✅ Imported opportunity #124 "AI for Tactical Edge Decision Support"
   Workspace: proposals/ai-tactical-edge-decision-support/
   Capture mode: Responsive (SBIR)

   Pre-populated from capture:
   • inputs/00_priority/what-matters.md          ← Go/No-Go + Next Steps
   • inputs/00_priority/solicitation-facts.md    ← Title, agency, NAICS, deadline
   • inputs/01_customer/customer-pocs.md         ← 2 POCs (1 from solicitation, 1 inferred)
   • inputs/03_teammates/suggested-partners.md   ← 3 AI-suggested primes
   • working/capture-intel.json                  ← Full raw JSON payload
   • working/proposal-brief.md                   ← Pre-filled brief with capture context

   Seeded from my-company/:
   • inputs/02_yourCompany/company-description.md
   • inputs/02_yourCompany/capabilities.md
   • inputs/02_yourCompany/contract-vehicles.md
   • inputs/02_yourCompany/past-performance.md

   Still needed from you:
   • Full solicitation PDF → inputs/00_priority/solicitation.md (or .pdf)
   • Any prior customer relationship context → inputs/01_customer/
   • Teammate commitments / LOAs → inputs/03_teammates/

   Next skill to run: /proposal-manager

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.