Recipes
7 ordered compositions of Tools that address one or more FailureModes. Recipes are the prescriptive flip-side of Incidents: patterns that have worked, or plausibly could, in practice.
context-hub (RAG grounding) → pydantic-ai (schema-validated outputs) → e2b (sandboxed execution) → pytest post-check. Fails fast on hallucinated symbols before any change leaves the sandbox.
case studiedpythonFabricationLogic errorcursor (IDE) → langfuse (trace / eval / dataset capture) → confident-ai-deepteam (scenario eval playground). Every prototype run emits structured traces the team reviews before promoting to...
anecdotalpolyglotContext pollutionLogic errorsourcegraph-cody (graph-indexed retrieval of the current API surface) → cursor (IDE generation) → tsc + eslint (post-generation).
case studiedtypescriptObsolescenceRed-team-first hardening
2 toolsconfident-ai-deepteam (adversarial scenarios) → guardrails-ai (runtime validators) → human review gate. Red-team scenarios double as runtime rules.
case studiedpolyglotSecurity vulnerabilityContext pollutionsourcegraph-cody (codebase graph) → tree-sitter (AST-level grep-before-write prompt rule) → semgrep (symbol-reuse rules). Forces the agent to discover and reuse existing functions rather than...
anecdotalpolyglotDependency blindnessSupply-chain-hardened agent
3 toolssocket (package-risk telemetry) → snyk (CVE scan) → daytona (sandboxed execution with signed MCP manifest). Intercepts hallucinated / malicious packages before they reach the dev machine.
case studiedpolyglotSupply-chain attackaider (commit-per-change makes test edits visible) → coverage-protected CI → semgrep rules on test-file diffs. Detects DAPLab DAP-07 silent-test-passing.
anecdotalpolyglotTest manipulation