MCP server · trust report

io.github.raintree-technology/docpull

2 checks that ran failed — supply chain and version behavior — so this release can’t be trusted as-is.

Publisher
io.github.raintree-technology
Repository
github.com/raintree-technology/docpull
Install
pypi:docpull, pypi:docpull, pypi:docpull
Versions
3
Inspected
5 of 6 checks · 75%
Scored
2026-09-01 · rubric 1.0.0

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

6 checks0 clean5 flagged1 not inspected
  • Injection surface

    unscannable25% of grade

    Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

    No tool descriptions, server instructions, prompts, or resources were fetched; nothing to scan for injection.

  • Supply chain

    fail25% of grade

    Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

    Python build script executes at build time: subprocess in build script

  • Credential hygiene

    warn15% of grade

    How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

    tests/test_project.py: assigns a literal to token (loca…et (len 12)) — non-provider secret literal; prefer environment intake. tests/test_provider_adapters.py: assigns a literal to api_key (test…ey (len 15)) — non-provider secret literal; prefer environment intake. tests/test_provider_adapters.py: assigns a literal to api_key (test…ey (len 12)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. src/docpull/context_packs/paper.py) — the recommended intake shape.

  • Permission scope

    warn15% of grade

    Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

    has both shell/exec and network egress — capable of download-and-run, though no such path was observed (src/docpull/context_packs/repo.py: subprocess.* call; src/docpull/auth_cli.py: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (src/docpull/context_packs/repo.py: subprocess.* call; src/docpull/hosted.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: src/docpull/context_packs/repo.py: subprocess.* call discloses shell/exec capability: src/docpull/context_packs/visuals.py: subprocess.* call discloses dynamic code execution: src/docpull/rendering.py: uses eval()/exec() discloses broad/destructive filesystem access: 4 files (src/docpull/hosted.py, src/docpull/mcp/tools.py, src/docpull/pipeline/steps/save_json.py, +1 more) Capability DISCLOSURE, not a verdict: static analysis sees the primitive is present and reachable, not whether its use is attacker-controlled.

  • Version behavior

    fail10% of grade

    Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

    vs prior 6.3.0 (minor bump): new capability primitive(s) in source: child_process, dynamic-eval, fs-write, http-client, subprocess; new hardcoded outbound host/BCC destination(s): 1.1.1.1, 10.0.0.1, 127.0.0.1, 169.254.169.254, 93.184.216.34, acme.test, api.anthropic.com, api.crossref.org, api.exa.ai, api.example.com, api.github.com, api.parallel.ai, api.tavily.com, arxiv.org, arxiv.org.evil.example, assets.example.com, blocked.example, blog.example.com, blog.python.org, books.example.com, books.toscrape.com, cdn.example.com, codeload.github.com, corp.proxy, datatracker.ietf.org, developer.apple.com, developer.hashicorp.com, developers.example, doc.rust-lang.org, docpull.raintree.technology, docs.anthropic.com, docs.aws.amazon.com, docs.exa.ai, docs.example.com, docs.foo.com, docs.parallel.ai, docs.python.org, docs.readme.io, docs.stripe.com, docs.tavily.com, docs2.example.com, doi.org, doi.org.evil.example, en.wikipedia.org, eutils.ncbi.nlm.nih.gov, evil.example, evil.example.com, evilsec.gov, example.com, example.invalid, export.arxiv.org, expressjs.com, fastapi.tiangolo.com, foo.readthedocs.io, github, github.com, github.com.evil.example, gitlab.com, hono.dev, hosted.example, jsonfeed.org, kubernetes.io, localhost., manual.example.com, news.example.com, nextjs.org, orm.drizzle.team, other.com, other.example.com, parallel.ai, pepy.tech, pkg.readthedocs.io, platform.claude.com, platform.example, platform.openai.com, platform.parallel.ai, proxy, proxy.example, public.example, pubmed.ncbi.nlm.nih.gov, pubmed.ncbi.nlm.nih.gov.evil.example, purl.org, pypi.org, python.langchain.com, raw.githubusercontent.com, react.dev, readme.io.evil.example, readthedocs.io.evil.example, registry.npmjs.org, schema.org, sec.gov, sec.gov.evil.test, sponsor.example.net, static.djangoproject.com, stripe.com, supabase.com, tailwindcss.com, user, vite.dev, widgets.example.com, www.apple.com, www.bbc.co.uk, www.bbc.com, www.djangoproject.com, www.postgresql.org, www.prisma.io, www.python.org, www.raindrop.ai, www.rfc-editor.org, www.sec.gov, www.sec.gov., www.sitemaps.org, www.tavily.com, www.w3.org, x, x.test.

  • Transport config

    warn10% of grade

    Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

    manifest/source references analytics behavior — disclosure

Version history

Each release plotted by grade against the safe line at B. A version that sinks below the line has lost its trusted standing — the shape of a rug-pull.

VersionPublishedGradeScoreInspectedChange
v6.5.0 · current2026-07-23 D455/6 · 75%B→D in v6.5.0
v6.3.02026-07-16 B1003/6 · 55%