proposal-storyboard
Skill4 starsChanged 3 months agoMITView on GitHub
Use this skill after proposal-solution-architect and before proposal-writer to create section-by-section storyboards that define evaluator questions, required answers, proof points, graphics, target length, and prohibited claims before prose drafting begins.
In this repository30 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adversarial-reviewSkill
- capture-demand-signalsSkill
- capture-intentSkill
- capture-portal-structureSkill
- capture-scorecardSkill
- capture-submissionSkill
- competitor-assessmentSkill
- compliance-checkSkill
- customer-intelSkill
- dashboardSkill
- evidence-checkSkill
- export-proposalSkill
- import-from-captureSkill
- narrative-spineSkill
- new-proposalSkill
- opportunity-quick-lookSkill
- past-performanceSkill
- pricing-analystSkill
- proposal-editorSkill
- proposal-graphicsSkill
- proposal-managerSkill
- proposal-patcherSkill
- proposal-solution-architectSkill
- proposal-writerSkill
- red-team-reviewSkill
- setup-companySkill
- statusSkill
- submission-summarySkill
- technical-reviewSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
17 rules. About 0 are common, 8 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 23
Run after /proposal-solution-architect and before /proposal-writer.
- Very specific: line 178
Read the existing working/storyboard.md first.
- Very specific: line 179
Preserve verbatim: every section whose Status is confirmed or edited — copy the block through unchanged, even if inputs changed.
- Distinctive: line 197
Every section must map to the question the evaluator is actually asking.
- Distinctive: line 220
If a claim is not supported, do not let the writer discover that during drafting.
- Distinctive: line 247
Every section must move the evaluator somewhere.
- Distinctive: line 256
If the answer is "the reader learns more about our capability," the section is not planned tightly enough.
- Very specific: line 260
Use reference/narrative-operating-modes.md to choose one mode for the whole response and, if needed, a section-specific override:
- Very specific: line 273
Create working/storyboard-coverage-map.md with a table:
- Distinctive: line 278
Use this to ensure the storyboard covers every known requirement before prose begins.
- Distinctive: line 280
If a requirement has no planned section home, mark it as a gap.
- Very specific: line 282
When working/evaluation-model.md exists, also add a factor coverage table — the Section-M analog of requirement coverage:
- Very specific: line 288
Every factor and pass-fail row from the evaluation model appears here.
- Very specific: line 293
Create working/storyboard-open-questions.md if any of the following are true:
- Very specific: line 310
After working/storyboard.md exists, /proposal-writer should treat it as the primary drafting plan.
- Distinctive: line 320
If the writer believes the storyboard is insufficient, it should flag the gap rather than writing around it.
- Distinctive: line 354
Do not draft proposal prose in this skill.
≈ “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
Proposal Storyboard Skill
Purpose
Decompose the approved narrative spine into a section-by-section writing plan before drafting prose.
The narrative spine (working/narrative-spine.md, produced by /narrative-spine) is the lengthwise argument — the case the whole proposal makes. The storyboard is the crosswise plan — what each section must do to advance that argument. It translates the approved spine, capture strategy, compliance matrix, solution architecture, and section pattern into a section-by-section writing plan, and prevents the proposal-writer from rambling, over-marketing, diving too deep technically, or forcing Shipley patterns where they do not improve the response.
The storyboard answers one question for every section:
What must the evaluator understand, believe, and score after reading this section?
When to Use
Run after /proposal-solution-architect and before /proposal-writer.
Recommended workflow:
/proposal-manager
/customer-intel
/competitor-assessment
/proposal-solution-architect
/proposal-storyboard
/proposal-writer
/proposal-editor
/compliance-check
/red-team-review --mode=redInputs
Always read
working/proposal-type.md— proposal type, page target, evaluator framing, section pattern.working/narrative-spine.md— the approved one-page argument (produced by/narrative-spine). This is the primary input. The storyboard decomposes the spine: each section storyboard should advance one movement of the argument, and a section that advances no movement is a candidate for cutting. Ifworking/narrative-spine.mddoes not exist, recommend the user run/narrative-spinefirst.working/compliance-matrix.md— requirement traceability and coverage targets.working/proposal-plan.md— evaluation criteria, win themes, bid strategy, discriminator candidates.working/solution-strategy.mdworking/architecture-concept.mdreference/section-patterns/<section_patterns>.md— derived fromworking/proposal-type.md.reference/narrative-operating-modes.md— select the prose mode that matchespage_target,submission_mechanism, andevaluator_framing. Record that mode in the storyboard header and apply it to section length, depth, and transition planning.working/evaluation-model.md— this solicitation's typed rubric (produced by/proposal-managerStep 4b viascripts/extract-evaluation-model.py). The source for each section'sTarget Evaluation Factorsfield. If it does not exist, recommend runningpython scripts/extract-evaluation-model.py --proposal <slug>first; proceed usingworking/proposal-plan.mdevaluation factors as the fallback key.
Read if relevant
working/requirement-matrix.md— for traceability columns in coverage map.working/capability-matrix.md— when storyboarding capability-mapped sections.working/assumptions-and-risks.md— when storyboarding risk/management sections.working/customer-profile.md— if/customer-intelwas run.working/competitor-assessment.md— if/competitor-assessmentwas run (required for ghosting angles).working/capture-intent.md— if/capture-intentwas run (drives belief objectives + prohibited claims).working/graphics-brief.md— if/proposal-graphicswas run.reference/proposal-conventions/<vehicle-id>.md— if convention file exists for this type.reference/doctrine/acquisition-guidance-catalog.md— apply its Applicability Matrix for thistype_idto decide which acquisition guidance a section must answer (drives the Acquisition Anchor field below). For DoD software efforts this is where the SWP MVP/MVCR cadence and, for Navy, RAISE 2.0 edge-ConMon become section-level discriminators.reference/editorial-voice-guide.md— for tone-section calibration.my-company/evidence-ledger.json— only if Phase C evidence is enabled.
Outputs
Write:
working/storyboard.md
working/storyboard-coverage-map.mdOptional, when useful:
working/storyboard-open-questions.mdStoryboard Standard
Each section storyboard must be specific enough that a proposal writer can draft from it without inventing strategy.
For each required section, produce:
## <section-id>: <Section Title>
**Status:** <proposed | confirmed | edited>
**Purpose:** <why this section exists>
**Evaluator Question:** <the question the evaluator is trying to answer>
**Required Answer:** <the plain-English answer the section must provide>
**Primary Takeaway:** <one sentence the evaluator should remember>
**Reader Movement:** <what the reader believes before this section -> what they should believe after it>
**Narrative Mode:** <compressed-brief | solution-brief | scored-volume | technical-merit | market-research | pricing-artifact>
**Transition Job:** <consequence | selection | proof | decision | none/formal response>
**Requirements Covered:** <Req IDs from compliance matrix, or "None/formal compliance not applicable">
**Target Evaluation Factors:** <EC-IDs + criterion text from working/evaluation-model.md that this section must win, or "⚠ NONE — no evaluation factor maps to this section">
**Evaluated Strength to Earn:** <the specific Strength or Significant Strength finding a mock evaluator should write after reading this section — one sentence, in S/W/D language>
**Acceptance Measure:** <the verifiable, countable commitment this section makes that an evaluator (or the Government at acceptance) could test — e.g., "onboard a model in ≤5 business days; two apps invoke ≥3 endpoints through one API"; or "None" for non-scoring sections. Calibration evidence (DIA DMA 2026-07): the acceptance-measure idiom was the single highest-leverage addition between initial cut and submission — the writer MUST emit this field's commitment in the section's prose.>
**Discriminator:** <why us vs. likely competitors for these factors, or "None">
**Acquisition Anchor:** <govt guidance this section must visibly answer, per the acquisition-guidance-catalog Applicability Matrix — e.g., "DoDI 5000.87: commit MVCR ≤12 mo + annual cadence", "RAISE 2.0: edge ConMon via stage-sync", "GAO-20-195G: state cost basis/method", or "None"; keep ATO/IL claims evidence-cited>
**Evaluation Factor / Scoring Tie:** <Section M factor, AOI, CSO criterion, SBIR merit factor, or reader concern — free-text context behind the Target Evaluation Factors above>
**Customer Language to Reuse:**
- <exact term or phrase from solicitation/customer docs>
- <exact term or phrase>
**Claims Allowed:**
- <claim that may be made, with evidence or source>
**Claims Prohibited or Must Be Qualified:**
- <claim that should not be made or must be softened>
**Proof Points / Evidence:**
- <evidence item, past performance, benchmark, deployment, named artifact, or CLAIM-UNSUPPORTED marker>
**Solution Content to Include:**
- <specific architecture/process/capability content>
**Competitor / Ghosting Angle:**
- <positive contrast only, or "None">
**Graphic Needed:** <Yes/No>
**Graphic Argument:** <what the graphic must prove, not merely show>
**Recommended Figure Caption:**
> *Figure X. <Title>.* <action caption asserting what the figure proves>.
**Target Length:** <word count or page fraction>
**Compression Rule:** <what to cut first if the section runs long, and what must survive>
**Do Not Say:**
- <phrase, claim, or rabbit hole to avoid>
**Drafting Notes:**
- <specific instructions to proposal-writer>Storyboard Quality Rules
Key Every Section to the Factors
For each section, decide which evaluation-model criteria (kind factor or pass-fail) the
section must win, and list them in Target Evaluation Factors by EC-ID. Map by genuine
subject-matter overlap between the factor's distinctive terms and the section's title plus its
mapped compliance rows — not by wishful assignment. Then:
- A section with no factor gets the warning verbatim:
⚠ NONE — no evaluation factor maps to this section. That warning is a strong "why is this section here?" signal — either the section earns its place another way (mandatory boilerplate, reader orientation) or it is a candidate for cutting. Say which inPurpose. - A factor with no section is a coverage gap — surface it in the coverage map's factor table (below), never silently.
Evaluated Strength to Earnstates the finding we want the mock evaluator to write; the Gold Team later checks the draft against exactly this sentence.
Regeneration Discipline (running log)
Storyboards are curated, not disposable. On every re-run of this skill:
- Read the existing
working/storyboard.mdfirst. - Preserve verbatim every section whose
Statusisconfirmedoredited— copy the block through unchanged, even if inputs changed. A human reviewed it; regeneration must not destroy that work. (Note inconsistencies with updated inputs inworking/storyboard-open-questions.mdinstead of editing the block.) - Regenerate only sections with
Status: proposed(or sections that are new). - New sections start at
Status: proposed. Humans flip status toconfirmed(right as generated) oredited(hand-tuned) as they review.
Be Specific
Bad:
Discuss technical approach.
Good:
Explain how local inference continues analyst workflows when SIPR/cloud connectivity is unavailable, then tie that architecture to the solicitation's DDIL requirement and the prior Navy/Army deployment evidence.
Answer Evaluator Questions
Every section must map to the question the evaluator is actually asking.
Examples:
| Section | Evaluator Question |
| Executive Summary | Why should I care, and why this team? |
| Technical Approach | Does this solution actually meet the requirement with manageable risk? |
| Management Approach | Can this team execute without creating schedule, staffing, or integration risk? |
| Past Performance | Has this team done similar work under similar conditions? |
| White Paper Approach | Is this credible enough to justify a meeting, pilot, CSO award, or follow-on ask? |
| RFI Response | Is this company relevant to the requirement and worth shaping around? |
Control Claims
The storyboard must distinguish between:
- verified claims
- reasonable inferences
- assumptions
- unsupported claims
- prohibited claims
If a claim is not supported, do not let the writer discover that during drafting. Flag it upfront.
Control Depth
Specify where technical detail belongs and where it does not.
Examples:
- Executive summary: operational outcome only; no container/runtime implementation detail
- Technical approach: architecture and interfaces; enough detail to establish feasibility
- Management approach: execution controls; no generic agile sermon
- White paper: pilot logic and transition path; avoid FAR-style boilerplate
Control Tone
The storyboard should state the intended tone for each section.
Examples:
- concise and executive-readable
- dense and evaluator-scored
- plainspoken and pilot-oriented
- technical but not academic
- factual and non-salesy
Plan Reader Movement
Every section must move the evaluator somewhere. Do not merely list the artifacts the writer should include.
For each section, answer:
- What does the reader believe before this section?
- What should the reader believe after this section?
- What proof, mechanism, or commitment creates that shift?
- What transition connects this section to the prior movement of the narrative spine?
If the answer is "the reader learns more about our capability," the section is not planned tightly enough. Reframe it as a decision, scoring, risk, or mission-value movement.
Select the Narrative Operating Mode
Use reference/narrative-operating-modes.md to choose one mode for the whole response and, if needed, a section-specific override:
compressed-brieffor 1-5 page white papers, short directed briefs, and tight web-form fields.solution-brieffor 5-10 page CSO/OTA/white-paper responses.scored-volumefor FAR, IDIQ, OTA full, and CSO full proposals.technical-meritfor SBIR and BAA technical merit responses.market-researchfor RFI and sources-sought.pricing-artifactfor ROM and price narratives.
The mode controls depth. A compressed brief should not spend paragraphs teaching the market. A scored volume should not hide compliance structure in elegant prose. A technical-merit response should not replace mechanism with sales language.
Coverage Map
Create working/storyboard-coverage-map.md with a table:
| Req ID / Eval Criterion | Storyboard Section | Planned Proof | Graphic | Risk / Gap |
Use this to ensure the storyboard covers every known requirement before prose begins.
If a requirement has no planned section home, mark it as a gap. Do not bury it.
When working/evaluation-model.md exists, also add a factor coverage table — the Section-M
analog of requirement coverage:
| EC-ID | Factor / Pass-Fail Criterion | Targeted By Section(s) | Gap? |
Every factor and pass-fail row from the evaluation model appears here. A factor targeted by
no section is flagged ⚠ GAP — we are not answering something we are scored on.
Open Questions
Create working/storyboard-open-questions.md if any of the following are true:
- a required claim lacks evidence
- a requirement has no clear section home
- a graphic is needed but the argument is unclear
- a technical commitment may exceed known capability
- the solicitation language is ambiguous
- page budget appears insufficient
- competitor ghosting would require unsupported assumptions
Format:
| Question | Why It Matters | Proposed Default if Unanswered |
Integration with Proposal Writer
After working/storyboard.md exists, /proposal-writer should treat it as the primary drafting plan.
The proposal writer should not invent:
- new win themes
- new discriminators
- new ghosting angles
- new major claims
- new graphics
- new section structure
If the writer believes the storyboard is insufficient, it should flag the gap rather than writing around it.
Anti-Patterns to Prevent
The storyboard must prevent:
- generic capability summaries
- executive summaries that read like company brochures
- technical sections that bury the answer under implementation detail
- sections that are individually compliant but do not advance the narrative spine
- transitions that announce structure instead of carrying the argument
- management sections filled with agile boilerplate
- repeated DDIL/cloud/edge explanations across every section
- proof points appearing after the evaluator needs them
- ghosting that sounds obvious or petty
- graphics that decorate instead of prove
- unsupported "leading provider" or "best-in-class" claims
Activity Trail
Append to working/activity.md:
## <timestamp> — proposal-storyboard — planned <N sections>, mapped <M requirements>, flagged <K open questions> → working/storyboard.mdAppend to working/ai-runs.jsonl:
{"schema_version":"ai-run.v1","timestamp":"<timestamp>","skill":"proposal-storyboard","proposal_id":"<proposal>","job_type":"storyboarding","provider":"anthropic","model":"claude-opus-4-7","input_tokens_estimate":null,"output_tokens_estimate":null,"cost_estimate_usd":null,"notes":"section-level storyboard and coverage map"}Final Rule
Do not draft proposal prose in this skill.
This skill plans the answer. The writer drafts the answer. The editor makes it sound human.
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.

