sextant
PYPI · SEXTANT-KB
Search your own PDFs, Markdown and text files from any assistant. Local, no key, cited passages.
Available components
Last scanned (15 hours ago)
Scores higher than 57% of PyPI servers
Category Search and Retrieval
How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →
Supply Chain Security74
- No malware found by supply-chain analysis.Pass
- A critical CVE affects chromadb, and no fixed version has been published yet. View diagnostics → Fail
- No install/post-install scripts declared.Pass
- 13 of 104 dependencies flagged as unhealthy. View diagnostics → Partial
Provenance & Transparency45
- Source repository is publicly reachable at the declared URL. View diagnostics → Pass
- Provenance check failed: no build-provenance attestation is published. See how to fix → View diagnostics → Fail
- Clear OSI-approved license (MIT).Pass
- Actively maintained (last published 4 days ago).Pass
- Disclosure check failed: no security disclosure policy was found in the source repository. See how to fix → Fail
Schema Quality & AI Usability70
- AI-judged instruction clarity (excellent).Pass
- Context-footprint check failed: tool/resource definitions use about 1331 tokens (~166/item across 8 items; 8 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management20
- Stability observed for 6 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage100
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 100% of tool parameters carry a description.Pass
- Structured output schemas are declared (100% of tools); any adoption earns full credit.Pass
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- We read all 8 captured tool definition(s), and no name or description among them implies an irreversible operation.Pass
- An AI judge read all 9 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a current MCP spec version (2026-07-28).Pass
Checks we ran (16)
- Malware scan
- Known CVEs
- Install scripts
- Dependency health
- Source repository
- Provenance
- License
- Maintenance
- Security disclosure
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Does not apply to this package (2)
- Endpoint SecurityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
- Transport & ReachabilityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 21 Nov 26 +16
- Malware scan: unverified → pass ▲ security
- 20 Nov 26 −15
- Malware scan: pass → unverified ▼ security
- 19 Nov 26 +16
- Malware scan: unverified → pass ▲ security
- 17 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 3 to 7. That category is still filling its 30-day observation window: 1 days of observed history at the previous scan, 2 at this one. The score rises as the window fills, whether or not the server changes.
- 16 Nov 26 0
- Malware scan: unverified → pass ▲ security
- Schema quality: 1056 → 1331 ▼ functional
- Stability: unverified → 0.03 ▲ functional
- Package version: 0.10.0 → 0.11.1 functional
- Package version: 0.10.0 → 0.11.0 functional
- 15 Nov 26 47
First indexed and scored.
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Analysed pypi/sextant-kb@0.11.1
Provenance No attestation
The registry publishes no build provenance for this version, so there is nothing to verify.
| Result | No attestation |
|---|---|
| Ecosystem | pypi |
| Reason | No attestation published |
Background: How many MCP packages publish verified provenance →
Vulnerabilities 8 findings
| ID | CVE | Severity | Vector | Fix available |
|---|---|---|---|---|
| GHSA-2wm9-hf6c-p5cr | CVE-2026-45830 | high | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H | no |
| GHSA-36p7-vc44-83pf | CVE-2026-45833 | critical | no | |
| GHSA-f4j7-r4q5-qw2c | CVE-2026-45829 | critical | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H | no |
| GHSA-xph7-9rjv-w5fr | CVE-2026-45831 | high | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H | no |
| PYSEC-2026-311 | CVE-2026-45829 | none | no | |
| PYSEC-2026-3813 | CVE-2026-45830 | high | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H | no |
| PYSEC-2026-3814 | CVE-2026-45833 | none | no | |
| PYSEC-2026-3815 | CVE-2026-45831 | high | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H | no |
Background: What a vulnerability scan can and cannot prove →
Dependencies 104 packages
| Packages resolved | 104 |
|---|---|
| Stale | 9 |
| No linked repository | 4 |
| Tree resolution | Complete |
Background: SBOMs and build attestations, explained →
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
kb_ann_benchmark ~145
Benchmark approximate-nearest-neighbour indexes over the current corpus. Builds hand-written HNSW and IVF-PQ indexes across a parameter sweep and measures each against exact search: recall@k, query-latency percentiles, build time and memory. This is an analysis tool for comparing index tradeoffs, not part of answering a question -- it does not retrieve anything for the user.
| Name | Type | Req | Description |
|---|---|---|---|
| grid | – | – | Optional sweep overrides: hnsw_ef_search, ivf_nprobe, ivf_m, etc. |
| k | integer | – | Neighbours per query to score recall over. |
| n_queries | integer | – | How many corpus vectors to hold out as queries. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_ann_compare ~131
Run one query through exact, HNSW and IVF-PQ search side by side and return each index's hits with its latency and which exact neighbours it missed. For inspecting how the approximate indexes behave on a specific question, not for answering one.
| Name | Type | Req | Description |
|---|---|---|---|
| hnsw | – | – | HNSW build/search params: m, ef_construction, ef_search. |
| ivfpq | – | – | IVF-PQ params: nlist, nprobe, m, nbits. |
| k | integer | – | Neighbours to retrieve from each index. |
| query | string | yes | The question or phrase to search with. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_ingest ~199
Add documents to the knowledge base so they become searchable. Each document needs 'content'; 'title', 'source', 'page' and 'category' are optional but 'source' and 'page' are what citations point at, so supply them when known. A caller that has already parsed a file may send a 'locators' list instead of 'page' -- entries of {kind, label, start, end} where kind is 'page', 'section' or 'table' -- and each chunk is then attributed to the region it starts in; a 'table' region is chunked one row at a time with its label (the header) prepended. Re-using an existing 'id' replaces that document. Long documents are split into overlapping chunks automatically. To index a file from disk, use the sextant-ingest command instead.
| Name | Type | Req | Description |
|---|---|---|---|
| documents | array | yes | Documents to store. Each requires a 'content' field. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_list ~88
List every document in the knowledge base with its title, category, size and what it is about: a model-written overview where one exists, and its opening lines either way. Use this for questions about the collection itself -- what the documents cover, which ones there are, which to read first -- rather than kb_search, which finds passages inside documents and has nothing to match a question about the whole.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
kb_search ~316
Search the user's private document collection for passages relevant to a question. Runs meaning-based and keyword-based retrieval together, so it handles both paraphrase and exact terms -- an error code, a surname, a section number -- but a full question still ranks better than one or two bare keywords. Returns passages with the source and page or section they came from, suitable for citation. Each result reports how it was scored: 'cross-encoder' scores are relevance in 0-1, where anything above about 0.5 is a genuine match. Returns an empty list when the collection holds nothing relevant, which means the answer is not in the knowledge base -- say so rather than guessing. A result flagged below_floor is the same verdict with the nearest passages attached for reading: they are ordered by embedding similarity, not judged relevant.
| Name | Type | Req | Description |
|---|---|---|---|
| limit | integer | – | Maximum passages to return. |
| min_score | number | – | Drop passages the reranker scores below this, 0.0-1.0. Leave at the default unless you are getting obvious noise; it is applied only when results are cross-encoder scored. |
| mode | string | – | Retrieval pipeline to run. Leave at 'rerank', the full one. The others are ablations for benchmarking: 'dense' is vectors only, 'lexical' is BM25 only, 'rrf' fuses both without reranking and returns… |
| query | string | yes | The question or descriptive phrase to search for. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_selftest_grade ~137
Step 2 of the library self-test. Takes the questions written for kb_selftest_sample's passages, each with its passage_id, searches every question exactly as kb_search does, and returns a short report: one score with how far chance could move it, the questions it missed, and the documents search cannot use (scanned, empty, unreadable or copies). Show the user the 'report' field exactly as it is; it already says the score is a rough guide.
| Name | Type | Req | Description |
|---|---|---|---|
| questions | array | yes | One entry per passage: {passage_id, question}, the passage_id exactly as kb_selftest_sample returned it. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_selftest_sample ~109
Step 1 of the library self-test, for when the user asks how well search works on their own documents. Returns a sample of passages spread across the library, with instructions: write one question per passage in your own words, then pass them all to kb_selftest_grade. Nothing leaves the user's machine except to you, the assistant already reading search results.
| Name | Type | Req | Description |
|---|---|---|---|
| seed | – | – | Repeat an earlier sample by passing the seed it reported. |
| size | integer | – | How many passages to test. |
Structured output declared, but exposes no named fields.
No examples provided.
kb_stats ~91
Report the state of the knowledge base: how many documents and chunks it holds, which embedding and reranking models back it, and where it is stored. Use this to check whether the collection is empty before concluding that a search found nothing. When the user chose a folder at install, 'folder' says whether it is still being indexed and lists any files that could not be read, with the reason for each.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
How do I install the sextant MCP server?
sextant runs locally as a PyPI package, launched with uvx sextant-kb. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
pypi · sextant-kb
claude mcp add hulagerushikesh-sextant -- uvx sextant-kb
What is the sextant MCP server?
sextant is an MCP server listed in the public MCP registry as io.github.hulagerushikesh/sextant. Search your own PDFs, Markdown and text files from any assistant. Local, no key, cited passages. This page covers its PyPI package (sextant-kb).
Is the sextant MCP server safe to use?
sextant scores 65 out of 100 on VerifyMCP. We recorded 8 known advisories against it as of 21 November 2026. It declares no install or post-install scripts. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the sextant MCP server expose?
sextant exposes 8 tools: kb_search, kb_ingest, kb_stats, kb_list, kb_ann_benchmark, and 3 more. Their descriptions and schemas cost roughly 1,216 tokens of context every time the server is loaded.
Is the sextant MCP server still maintained?
sextant is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What licence is the sextant MCP server under?
sextant declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.
What did VerifyMCP test on the sextant MCP server?
VerifyMCP ran 16 checks on sextant, covering Malware scan, Known CVEs, Install scripts, Dependency health and 12 more. Endpoint Security and Transport & Reachability do not apply to a PyPI package, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.