Code Search CARE Agent
v1.0.0activeScientific code repository discovery agent that searches NASA-verified repositories, Science Discovery Engine (SDE), and ADS to find publicly available code repositories relevant to scientific and technical tasks. Outputs are delivered via a structured schema and interactive chat with the user for clarification, guidance, approval gates, or status updates.
by NASA-IMPACT akd-ext contributors (NASA-IMPACT) · other · search-retrieval
T1 · Conformantwhat's this?
tested on
gpt-5.2
license
Apache-2.0
framework
openai-agents-sdk
citable url
https://agentarium.science/a/code-search-care/v/1.0.0
Guardrails & validationexplicitly declared by the author — shown so you can judge, not registry-verified
Guardrails declared — author-stated
✓
Non-prescriptive outputs
Never recommends, selects, or endorses repositories — produces comparative, ranked descriptions only.
✓
Prohibited language filter
Must not use language such as 'best', 'recommended', 'final choice', 'approved', or 'use this'.
✓
Read-only operation
No execution, cloning, downloading, testing, or code generation permitted in any step.
✓
Candidate lifecycle tracking
Running Candidate List maintained across all discovery steps; candidates removed only at Step 7 with documented reasons.
✓
No fabrication
No fabricated repositories, metadata, or capabilities; missing information reported as unknown.
✓
No credentialed sources
Never accesses private, gated, or credential-restricted repositories or sources.
✓
Disclosure of excluded and unlocated codes
All candidates found but not ranked, and all expected codes not located, must appear in dedicated output sections with explicit reasons.
✓
Abstain-only-when-empty rule
Returns zero results only when no discovery channel yields any plausible candidate; never abstains if candidates were found.
✓
Context budget
Target no more than 10 total tool calls across the full pipeline to avoid context exhaustion.
✓
Dual-use caution
Dual-use or sensitive domains surfaced only with explicit caution; no implication of legality or fitness for use.
Validation methodology — author-stated
Tested50 known Earth-science queries with ground-truth CMR concept IDs (seed-record placeholder).
DataCurated query set from NASA-IMPACT teams (seed-record placeholder).
MetricReference collection appears in ranked top-5 (seed-record placeholder).
ResultSeed record — author should publish a real validation before public release.
CaveatThis is a registry seed record; the validation block was filled with placeholders. The author is encouraged to submit a new version with real numbers and a real caveat.
Required tools — live healthlive status of MCP endpoints this agent depends on; not registry-verified
cmr_mcp_serverapproval: never
healthy
NASA Common Metadata Repository (CMR) MCP server. Read-only Earth-observation catalog search.
allowed: sde_search_tool, repository_search_tool, code_signals_search_tool
ads_searchapproval: never
missing from registry
allowed: ads_search_tool, ads_links_resolver_tool
web_search_toolapproval: never
missing from registry
allowed:
Reproductionsindependent runs by other scientists — Tier 5 trigger
Ran this agent yourself? File an independent reproduction — it can move the listing to Tier 5.Sign in to reproduce
Other disclosuresas described by the author
- Intended use
- Helps scientists and engineers discover publicly available code repositories relevant to a stated scientific or technical task. Built for exploratory, read-only, human-in- the-loop code discovery — the user retains control over final code selection and fitness-for-purpose assessment. Covers NASA and community scientific software across Astrophysics, Earth Science, Heliophysics, Planetary Science, and related domains.
- Out of scope
- Does not recommend, select, endorse, or judge repositories for fitness of use. Does not assess code quality, correctness, safety, certification, or legality. Does not execute, clone, download, or test code. Does not access private, gated, or credential-restricted repositories. Does not generate code. Does not operate on queries with no plausible candidates when zero channels return results.
- Known failure modes
- Expected Codes checklist may be incomplete for niche subfields. ASCL "Code site" URLs may be outdated if ASCL has not been updated recently. ADS query depth is capped at 5 rows per query, so low-citation codes may be missed. Web search step is limited to 3 queries — missing codes in that step are documented but not re-tried. Non-Astrophysics queries skip the ADS step entirely and rely solely on SDE and web search for enrichment, which may miss community codes not in NASA repositories.
Worked example
input
Provide an NPM module for accessing Firefly API to get and visualize astronomical archival data.
output
## Ranked Repositories ### 1. firefly-api-client - **Primary URL:** [https://github.com/Caltech-IPAC/firefly](https://github.com/Caltech-IPAC/firefly) - **Secondary URL:** — - **Rationale for inclusion:** Found via NASA Repository Search; official Caltech IPAC Firefly visualization toolkit with REST/WebSocket API for astronomical archival data access. - **Fit notes & limitations:** JavaScript/TypeScript client; covers FITS visualization, catalog overlays, and archive queries. Not a standalone NPM module — client integration requires a running Firefly server instance. - **Provenance:** NASA Repository Search - **ADS Evidence:** - Bibcodes: 2019ASPC..523..521R - Citation count: 12 - Usage summary: Referenced in papers describing IRSA archive visualization workflows. ## Search Notes - Candidate list based on NASA Repository Search (2 queries) and SDE enrichment. - ADS step applied (Astrophysics domain). Citation counts used as ranking signal only. - No candidates were excluded; no expected codes were unlocated. - Assumption: user requires a client library rather than a server deployment.
What this listing is. A structured, format-conformant submission, screened for topic and obvious safety issues. The registry does not verify that the agent is correct, that it works, or that the author's disclosures are accurate. Evaluate before relying on it for research.