More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    SBIR/STTR Proposal Pack

    1

    Bid decisions, proposals, and budgets for SBIR/STTR topics, with live DoD topic Q&A and a search of every past award on sbir.gov.

    $49

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    SBIR/STTR Proposal Pack

    SBIR/STTR Proposal Pack

    Example session with this skill installed

    Should we bid on Air Force SBIR topic DAF26BX06-DP027? We're Halyard Sensing, 11 people, all US citizens, founder-owned. We make passive drone-detection sensor nodes (thermal camera + microphone array). We've built 6 nodes, about $7,500 each in parts. Field tested last spring at a private airfield: detected quadcopters with radios off out to about 180 m and up to 200 m altitude, outputs a track with bearing and the node's GPS position. No SBIR awards before. I don't have time to answer a bunch of questions, just give me your read with what you have.

    • Read your context and instructions
    • Compiled the sbir sttr proposal

    Real session, run 2026-10-04 with Claude Sonnet against a live Air Force topic. The topic, rules, Q&A, and award data are real. The company is made up for the test.

    DAF26BX06-DP027: bid read for Halyard Sensing (2026-10-04)

    Read: no-bid as things stand. Bid only if you can pass the cost gate below by the end of Wednesday, Oct 7.

    The deciding number is cost. Your hardware fits the topic. Your price doesn't.

    1. The cost math (the deciding requirement)

    The topic says the

    final system must cost under $100K at full-rate production. Q11 confirms this is the whole multi-node system, not one node. Q7 says it covers "every fielded component (software and hardware)." Only shore-power consumption is excluded.

    The system must also cover a perimeter of at least 10 km (Q9: any polygon, not a straight line). Each node must detect out to at least 100 m horizontally.

    Node coverage assumedSpacing along the lineNodes for 10 kmYour parts at $7,500/nodeLeft per node for a $100K system
    100 m (topic minimum)~200 m~50$375,000$2,000
    180 m (your field result)~360 m~28$210,000~$3,570

    Parts alone are 2x to 3.75x the whole-system ceiling. That is before compute at scale, comms, mounts, power, install, C2 and API software, licenses, margin, and any overlap you need for track handoff.

    Production pricing will not close a gap this size. You would need roughly a 3x to 6x all-in cost cut, depending on how much node overlap you need.

    Q1 is still unanswered. It asks whether installation, backbone, one-time licenses, and production quantity count toward the $100K. The answer could move the math, and probably not in your favor.

    Upside you can use. Q2 allows mixed sensor payloads across nodes. A cheap single-modality node for most of the line, with thermal+acoustic only at hard spots, is a legitimate design. You would have to show it still detects non-RF drones across the whole line.

    2. Fit against the topic

    RequirementYour evidenceRating
    Detect and track non-RF Group 1-3 UAS at low altitude (the D2P2 gate)Quadcopters, radios off, ~180 m, up to 200 m altitude, at a private airfieldPartial. The Group 1 test is on point. Nothing on Group 2 or 3 (larger and faster).
    Output basic track plus node locationBearing plus node GPSStrong. Q11 says the topic doesn't prescribe track fields.
    Detect from 1 m up to 300 m (500 m objective)200 m testedGap. 100 m short of threshold.
    At least 100 m horizontal per node~180 mStrong, assuming 180 m is ground range and not slant range. Check this.
    Operate in clutter and tell drone from not-a-droneNot statedUnknown. An airfield is open and low-clutter.
    10 km boundary, nodes networked6 nodes builtGap. Not required at the feasibility stage (Q3, Q11), but the proposal must address it.
    Cost under $100K for the whole system$7,500 per node in partsGap, and it decides the bid (section 1).
    Remote node control, full external API, Medusa C2 or other AF C2Not statedGap. Q8 says there is no pre-award access to Medusa docs.
    All-weather operation, cyber-secured, CMMC Level 2 (self)Not stated. Tests were in spring only.Gap. Q9 requires all weather drones can fly in, and cyber security. Thermal and acoustic both degrade in rain and wind.
    Heading of targetYou output bearingPartial. Bearing is not heading. Q11 says no fields are prescribed. Show how you get heading from the track.

    Four rows are gaps and three are partial or unknown. That is well over a third. The Phase I-type feasibility rows are your strongest, which is why this is a close call and not an easy no.

    3. Eligibility and the feasibility gate

    Size, citizenship, ownership. 11 people, all US citizens, founder-owned. That is a clean profile on the facts given. Final eligibility calls are yours and counsel's.

    • No prior SBIR. Not a barrier. D2P2 does not require a Phase I.

    Your field test counts if it was not federally funded. The feasibility work cannot come from or extend any SBIR/STTR work. It must have been done mainly by your company and PI. Yours was self-funded, so this fits.

    Disqualification risk. The instructions say a D2P2 proposal is disqualified and not evaluated if the Phase I-equivalent documentation doesn't establish feasibility. The Government will not tell you in Q&A whether your prototype qualifies (Q8, Q11). A written test report is the whole gate. You need: test plan, targets used (make, model, weight), confirmation the radios were off, ranges, detection and false-alarm counts, track plots.

    • Rules to plan around.
      • 35 pages, 11-pt font, 1-inch margins.
      • One proposal per firm for this open topic.
      • The PI must work for you more than half time.
      • FWA training must be completed before the proposal can be submitted.
      • Volume 7 foreign-affiliation disclosures are required. They are signed under penalty of perjury, and the Air Force screens for foreign ties.
      • Technical, cost, commercialization report, and supporting volumes all go in through DSIP.

    Not confirmed. I have not seen SAM.gov/UEI, SBA registry, or DSIP status, or who your PI is. These take time. Check them now.

    4. Who already won nearby (public sbir.gov award file, searched today)

    No award shows under this topic code yet. It is a new release. Close neighbors from the last two years:

    Sphere Optics: Air Force D2P2, "Optical Fence Line with Spherical UAS Detect, ID and Track," $1.8M (2025). This is the same topic family and the same fence-line framing, and the Air Force has already paid for an optical version.

    PicoGrid: Army, "Low-Altitude Passive Integrated Sensors (LAPIS)," $2.0M Phase II (2025). It uses passive RF sensing and a tethered UAS, with C2 integration (TAK and others).

    Physical Sciences Inc.: Army, "VIS/LWIR Event Based Sensor Suite for Passive UAS Detection," $2.0M Phase II (2025). Passive visible and thermal sensing from a 259-person firm.

    • Anno.AI: Air Force, "EchoGuard passive drone detection platform," Phase I (2025).
    • Hidden Level: Air Force Phase II, $1.6M (2024), for subscription airspace monitoring.

    Nokomis: passive RF detection, $1.2M to $1.9M Phase IIs (2025). This is RF-based, so a different approach from yours.

    Positioning. The area is crowded with larger, better-funded firms that already hold Phase IIs. A passive thermal+acoustic node with a defined cost path is a distinct pitch only if the cost holds. Without that, the evaluators have no reason to pick you over a firm with an existing award and an integrated C2 story.

    5. How the proposal would be scored

    • Factors A (Defense Need) and B (Technical Approach) are equal. C (Commercialization) is lower. All non-cost factors together matter significantly more than cost, but "where technical evaluations are essentially equal, cost and/or price" decides.

    Defense Need. Signed letters of support from Government personnel are called out as indicators. You have none mentioned, so get at least one if you bid.

    • Award. $2M over 24 months. Multiple awards are possible.

    Where you are weakest. Technical Approach has to answer cost, 300 m altitude, all weather, C2 and API, and the 10 km network, all in 35 pages. Your current evidence covers only one of those.

    6. What to do next

    1. Today or tomorrow (Q&A closes Oct 7, 12:00 ET): submit questions in DSIP.
      • Does the $100K include installation, backbone, one-time licenses, and what production quantity does "full-rate production" assume? Q1 is still open. Add a vote or follow-up if you can.
      • Is the 100 m node range a ground range? What altitude band must each node cover at that range?

    By Oct 7: cost gate. Build a bottom-up, volume-priced BOM for a node, plus software, comms, install, and C2 integration. Divide $100K by a node count that gives 10 km at your demonstrated range. If the all-in per-node figure is not near $3K, stop.
    3.

    Same days: evidence gate. Check whether your test data can be written up as a feasibility report. Check whether you have any Group 2/3 or fixed-wing data, any rain, wind, or clutter data, and anything above 200 m altitude.
    4.

    If both gates pass: register in DSIP, finish FWA training, and line up one Government letter of support. The deadline is Oct 21 at 12:00 ET, 17 days from today. Draft the feasibility report first, because it can disqualify you.
    5.

    If either gate fails: skip this one. Run the same check on the next DAF open topic that fits a $7,500 node better, or one that does not carry a $100K whole-system cap.

    7. Assumptions I made because you didn't have time for questions

    • "180 m" is horizontal range and "200 m altitude" is the highest altitude you tested, not a slant range.
    • $7,500 is parts only, with no labor or margin.
    • Tests used Group 1 quadcopters only.
    • Nobody on staff has a Medusa or Air Force C2 integration history.

    If any of these is wrong in your favor, the read improves. The cost math is the one most worth checking.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Write SBIR and STTR proposals from the actual topic, and find out whether to bid before you spend three weeks on one.

    What it does

    One skill with six workflows for small businesses responding to DoD, NIH, NSF, DOE, and NASA topics:

    1. Break the topic into what reviewers score. Evaluation criteria with their weights, each requirement with the question a reviewer will ask, and a fit rating against what your company has.
    2. Find who already won. Searches the full public sbir.gov award file on your machine (about 395 MB, downloaded once). Shows prior winners, their approaches, and who turns Phase I awards into Phase II.
    3. Draft the technical volume in the agency's required section order, with a phrase-by-phrase check of your feasibility proof for a Direct to Phase II.
    4. Plan the Phase II and Phase III path: customer, buying path, evidence of demand, the commercialization questions the solicitation requires, and a results schedule.
    5. Check eligibility and registrations: ownership, PI employment, work split, subcontractor documents, supplier and foreign-risk questions, and each registration with the date it has to be started by.
    6. Build the budget and cost volume. budget_check.py adds your fee and checks the price against the award ceiling, the small business work share, and the cap on outside spending.

    It pulls DoD topics for you

    dsip.py fetches any DoD topic from DSIP: the topic record with deadlines and times, every published Q&A, and the instructions PDF. It prints the lines that set that topic's page limit, award ceiling, and evaluation weights, maps the component's section with a line number for each heading, and saves the topic's official text.

    Why the details matter

    The demo below is a real run on a 2026 Air Force topic. The topic says the system must cost under $100K. The published Q&A says that ceiling covers the whole 10 km system, not each sensor node. In one test run, Claude Sonnet with web search but without this pack said to bid. With the pack, it read the Q&A, did the arithmetic, and said no-bid.

    A second run on the same topic caught a different problem. A $1.95 million budget looked safe under the $2 million ceiling. The Air Force requires a fee in a Phase II price, and at 7% the price went $86,500 over. Proposals over the ceiling are not evaluated.

    Current as of October 2026

    • DoD's 2026 SBIR Commercial Solutions Opening, read from the official text: Volume 7 foreign disclosures, a CMMC level on each topic, component rules that override the program-wide ones, and the Air Force's full Direct to Phase II requirements.
    • The April 2026 reauthorization and what to look for in a solicitation because of it.
    • Every reference file carries the date it was last checked and its sources.

    What's in the zip

    SKILL.md, 6 workflows, 5 reference files, 6 templates, a worked example on a real Air Force topic, and three scripts (dsip.py, sbir_awards.py, budget_check.py). Python 3 standard library only.

    Limits

    It doesn't invent technical results, rates, or prices. Where your evidence is missing it leaves a marked gap. Eligibility questions about affiliation and ownership are flagged for you and counsel.

    By Daniel Kinney-Spears. Twelve years writing federal proposals, most recently running capture and proposal at Anduril, and before that at AWS.

    Deepest on DoD: the topic fetcher and the component rules file are DoD-specific, and the fullest rules are the Air Force's Direct to Phase II. For NIH, NSF, DOE, and NASA, page limits and section orders are typical values to confirm against the solicitation. It doesn't invent technical results, rates, or prices; missing evidence is left as a marked gap for you to fill. Ownership and affiliation questions are flagged for you and counsel, not decided. Rules were checked in October 2026 and change every cycle, so the solicitation always governs.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Requires

    Sbir
    Dodsbirsttr
    Python

    Permissions

    Terminal / Shell
    Network Access
    Read Files
    Write Files

    Allowed Hosts

    data.www.sbir.gov
    www.dodsbirsttr.mil

    File Scopes

    pursuits/**
    ~/.cache/sbir-awards/**

    The scripts download two public data sources: the sbir.gov award file and DoD topic documents from dodsbirsttr.mil. They save them to a cache folder and to your proposal workspace. They upload nothing. The skill reads the solicitation files you give it and writes its outputs as markdown in your workspace.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    The pack includes six automated workflows for topic analysis, award history searching, drafting technical volumes, commercialization planning, eligibility verification, and budget auditing. It also provides specialized Python scripts to fetch live DoD Q&A and validate cost volumes against agency-specific caps.

    The 'dsip.py' script fetches real-time topic data, deadlines, and official Q&A directly from the Defense SBIR/STTR Innovation Portal (DSIP). This ensures your AI analyzes the most current instructions, including late-breaking clarifications that web searches often miss.

    The budget workflow uses 'budget_check.py' to verify your fee, small business work-share percentages, and outside spending caps against the specific solicitation ceiling. It is designed to prevent 'hard-stop' disqualifications, such as exceeding a price cap due to an overlooked fee calculation.

    The skill includes a local 'sbir_awards.py' script that searches a comprehensive 395 MB database of past sbir.gov awards. This allows you to identify previous winners, study their technical approaches, and determine which companies successfully transition from Phase I to Phase II.

    The pack requires a standard Python 3 environment and is optimized for use with advanced LLMs like Claude 3.5 Sonnet or GPT-4o. It uses the Python standard library only, ensuring compatibility across most systems without requiring complex third-party library installations.

    All reference files and workflows are current as of October 2026, incorporating the latest SBIR reauthorization requirements, CMMC disclosure rules, and agency-specific Direct to Phase II mandates. Each file includes a timestamp and source log for auditability.

    New in Legal & Compliance