Skip to content

BdM-15/capture-insights / skills/oci-sweeper

oci-sweeper

Verified

Federal Organizational Conflict of Interest (OCI) sweeper backed by FAR Subpart 9.5 (9.501-9.508) and the active Theseus workspace knowledge graph. USE WHEN the user asks about OCI risk on a bid, organizational conflicts, incumbent conflicts, biased ground rules, unequal access to information, impaired objectivity, or any pre-bid OCI due diligence. Pulls `company`, `incumbent`, `subcontractor`, `customer`, `program_office`, and prior-contract relationships from the workspace KG, classifies each potential conflict into one of the three FAR 9.505 classes (biased ground rules, unequal access, impaired objectivity), and emits a structured findings envelope with mitigation recommendations (firewall, NDA, recusal, novation). DO NOT USE FOR FAR clause coverage audit (use `compliance-auditor`), proposal prose (use `proposal-generator`), competitor research (use `competitive-intel`), or pricing (use `price-to-win`).

evals/evals.json

View on GitHub
{
  "skill_name": "oci-sweeper",
  "schema_version": "1.0",
  "purpose": "Smoke-tests that oci-sweeper invokes the right KG queries, references the correct FAR Subpart 9.5 taxonomy, and produces a structured OCI sweep envelope with per-finding mitigation classes.",
  "evals": [
    {
      "id": "oci-001-incumbent-impaired-objectivity",
      "prompt": "Scan this workspace for OCI risk. The incumbent on the predecessor contract is one of our partners on this bid.",
      "expected_signals": [
        "FAR 9.505-3",
        "impaired objectivity",
        "incumbent",
        "prior contract",
        "mitigation",
        "firewall"
      ],
      "anti_signals": ["no OCI risk found", "out of scope"],
      "expectations": [
        "calls kg_entities with types including 'company' or 'incumbent' or 'subcontractor'",
        "calls kg_query for relationships involving partner / incumbent entities",
        "envelope contains findings[] with at least one finding tagged class='impaired_objectivity'",
        "each finding has a mitigation_recommendation referencing FAR 9.504 or 9.505",
        "envelope contains coverage_summary with counts for unequal_access / impaired_objectivity / biased_ground_rules"
      ]
    },
    {
      "id": "oci-002-customer-side-personnel-conflict",
      "prompt": "Sweep for OCI: one of the customer's KO support contractors is also a teaming partner we listed.",
      "expected_signals": [
        "FAR 9.505-1",
        "biased ground rules",
        "customer",
        "support contractor",
        "mitigation",
        "non-disclosure"
      ],
      "anti_signals": ["no conflict", "skill not applicable"],
      "expectations": [
        "envelope flags biased_ground_rules class for customer-support / partner overlap",
        "finding cites FAR 9.505-1 OR 9.505-2 in citation field",
        "mitigation includes 'firewall' OR 'recusal' OR 'non-disclosure'",
        "no fabricated FAR sections (only Subpart 9.5 references)"
      ]
    },
    {
      "id": "oci-003-unequal-access-data-from-prior-work",
      "prompt": "Did our prior work on a related agency contract give us unequal access to non-public information that biases this bid?",
      "expected_signals": [
        "FAR 9.505-4",
        "unequal access",
        "non-public information",
        "prior contract",
        "mitigation"
      ],
      "anti_signals": ["no risk", "not enough data"],
      "expectations": [
        "envelope contains at least one finding with class='unequal_access'",
        "finding references FAR 9.505-4 in citation field",
        "mitigation_recommendation includes 'NDA' OR 'data segregation' OR 'firewall'",
        "envelope has overall_risk_level field set to one of: low, medium, high",
        "every finding has a chunk_id OR entity_name traceability field (no ungrounded claims)"
      ]
    },
    {
      "id": "oci-sweeper-grounding-floor",
      "name": "Grounding-audit floor",
      "prompt": "Scan this workspace for OCI risk. The incumbent on the predecessor contract is one of our partners on this bid.",
      "expectations": [
        "tools/audit_skill_grounding.py --skill oci-sweeper reports grounding_ratio >= 0.90 (OCI sweep is FAR-citation-heavy; the on-disk JSON envelope source_chunks are the citation-of-record, the chat narrative is the human-readable digest).",
        "tools/audit_skill_grounding.py reports claims_exempt is at least 0.40 of (claims_total + claims_exempt).",
        "No regression below 0.7 across at least three consecutive runs of the same prompt."
      ],
      "expected_signals": [
        "tools/audit_skill_grounding.py --skill oci-sweeper reports grounding_ratio >= 0.90 (OCI sweep is FAR-citation-heavy; the on-disk JSON envelope source_chunks are the citation-of-record, the chat narrative is the human-readable digest).",
        "tools/audit_skill_grounding.py reports claims_exempt is at least 0.40 of (claims_total + claims_exempt).",
        "No regression below 0.7 across at least three consecutive runs of the same prompt."
      ],
      "anti_signals": [
        "grounding_ratio drops below the per-skill floor of 0.90",
        "the skill emits OCI findings without citing the FAR 9.5 sub-clause that triggered the class",
        "the skill confuses the three OCI classes (biased ground rules vs. unequal access vs. impaired objectivity)"
      ]
    }
  ]
}
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall