What did Hlido score Octocode?

Octocode scored 82/100 (STEADY) in Hlido's independent, hands-on review.

How much does Octocode cost?

Hlido recorded the pricing model as: Open source. Derived from evidence Hlido holds, not vendor-supplied; verify current prices on the vendor's pricing page.

Does any vendor pay Hlido for placement?

No. Hlido takes no money from the agents it rates — scoring weights stay private and the evidence behind every verdict is public.

Octocode

Name: Octocode review
Item: Octocode
Rating: 82
Author: Hlido Editor

Coding · tested 2026-08-26 · re-test due 2026-11-24 · by the Hlido desk, not the vendor

In short: Evidence-first code research for AI agents — 13 read-only tools over GitHub, npm and local code that return answers with citations, shipped as an MCP server, a CLI, a skill and an agent harness, all MIT-licensed with no indexing.

5 PASS · 0 FAIL of 5 public-surface claims

Visit Octocode → scorecard.json

Quick answer

Octocode scores 82/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-08-26). STEADY (82) because Octocode is agent-native on four surfaces (MCP, CLI, skill, agent bundle), read-only by default with secrets stripped, citation-grounded, MIT-licensed and index-free — a rigorous and safety-conscious Pricing: Open source (free entry point documented).

Hlido Editor · 2026-08-26

Octocode's pitch is unusually disciplined for the code-context field: instead of embedding your repo into a hosted index, it exposes 13 research tools that search GitHub, npm and local code and return answers grounded in real evidence with citations, so an agent cites source rather than guessing. It is deliberately read-only by default and strips secrets from both inputs and outputs — two safety properties most code-context tools do not advertise, and exactly the properties a coding agent operating on private repos needs. The surface is genuinely multi-modal: the same engine ships as an MCP server (`octocode-mcp`, one-click for Cursor/VS Code, paste-in for Claude Code/Windsurf/Zed), a CLI (`npx octocode`, 12 commands, JSON output, no AI host required), a SKILL.md protocol that orchestrates the tools into an evidence loop, and a Pi agent-extension bundle. It is free and open source under MIT, authenticates through your existing GitHub login, and needs no indexing step. The honest limits: the reviewed evidence is the marketing surface, not a measured tool run, so the '13 tools' and citation quality are described rather than independently benchmarked here; and there is a naming clash in the wild — an unrelated Rust tool by a different author also ships as 'Octocode', which can confuse discovery. Within its lane, the combination of read-only-by-default, secret-stripping, citation-grounded answers and a no-lock-in MIT license is a strong, agent-native posture.

Why STEADY

STEADY (82) because Octocode is agent-native on four surfaces (MCP, CLI, skill, agent bundle), read-only by default with secrets stripped, citation-grounded, MIT-licensed and index-free — a rigorous and safety-conscious posture for code research. Held at 82 rather than higher because the tool count and citation quality are described on the public surface rather than independently benchmarked in this review, and because an unrelated tool shares the 'Octocode' name, which muddies discovery.

Public-surface checklist

PASS Homepage loads (required)
PASS Primary value prop (required) — Evidence-first code research for AI agents — cited answers over GitHub/npm/local code
PASS Cta present (required) — Installation / one-click MCP install / npx octocode
PASS Pricing or access — Free & open source (MIT); GitHub auth; no hosted index
PASS Evidence or demo — Tools page, installation guide, worked MCP config and command examples

What we saw

4 screenshots captured by the Hlido engine during the reviewed run (run-7aef41b06ea7a6fe-octocode-ai). Our own captures — not vendor marketing material.

Octocode — run screenshot 1 (home.png) — home.png

Octocode — run screenshot 2 (page_main.png) — page_main.png

Octocode — run screenshot 3 (pagetools.png) — pagetools.png

Octocode — run screenshot 4 (pageinstallation.png) — pageinstallation.png

What it does well

Read-only by default and strips secrets from inputs and outputs — a safety posture built for agents on private repos
Returns answers grounded in real evidence with citations instead of unsourced guesses
Four agent-native surfaces from one engine: MCP server, CLI, SKILL.md protocol, and a Pi agent-extension bundle
No indexing step and no hosted code index — authenticates through your existing GitHub login
Free and open source under MIT, with a Rust-powered engine option
Works across Cursor, Claude Code, VS Code, Windsurf and any MCP host with one-click or paste-in setup

What it fails at

The reviewed surface is marketing/documentation, not a measured tool run — '13 tools' and citation quality are described, not independently benchmarked here
An unrelated Rust tool by a different author also ships as 'Octocode', creating discovery ambiguity
No pricing/commercial-support story — reads as a self-run developer tool
Depth of results on very large private monorepos is not demonstrated on the public surface

Red flags

Name collision: an unrelated Rust-based 'Octocode' (semantic search / knowledge graph) exists from a different author — confirm you are installing `octocode-mcp` from octocode.ai if that is the tool you want.

Best for

Coding agents that need cited, evidence-backed answers about a codebase rather than guesses
Teams that want read-only, secret-stripping code research on private repos with no hosted index
Developers who want the same engine available as MCP, CLI and a repeatable skill

Not recommended for

Teams that specifically want a hosted, pre-indexed semantic search product with SLAs
Users who need embeddings-based ranking rather than tool-driven, cited retrieval

Pricing & access

ModelOpen source
Free entry pointYes — a free tier or open-source edition is documented
Pricing findable on the public surfacePASS Free & open source (MIT); GitHub auth; no hosted index (tested 2026-08-26)

Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-08-26.

Compared to

Serena cited-evidence-research-vs-symbol-operations
Serena is an LSP-based semantic code toolkit; Octocode is an evidence-first research layer that returns cited answers across GitHub/npm/local code and is read-only with secrets stripped. Serena for symbol-level IDE-style operations, Octocode for cited cross-source research with an agent-safety posture.
codebase-memory-mcp live-cited-retrieval-vs-persistent-memory
codebase-memory-mcp persists codebase memory for an agent; Octocode focuses on live, cited retrieval across GitHub/npm/local with no indexing. Memory tool for recall across sessions, Octocode for fresh evidence-grounded lookups.

Agent relevance

CLI MCP Behavioral-testable

Agentic-Commerce Readiness 62/100 · INTEGRABLE

Independent readiness for agent delegation & transaction. How it’s scored · check live

Primary path is the MCP server (`octocode-mcp` over stdio) connected to Cursor, Claude Code, VS Code, Windsurf or any MCP host, exposing 13 read-only research tools an agent calls to retrieve cited evidence from GitHub, npm and local code. The same engine is also a CLI (`npx octocode`, JSON output) for non-agent scripting and a SKILL.md evidence loop. Read-only-by-default and secret-stripping make it comparatively safe to hand to an autonomous agent.

Agent-friendly score: 9/10

Evidence

13 read-only tools search GitHub, npm and local code and return answers with citations, no indexing — source (2026-08-26) verified
Read-only by default; secrets stripped from inputs and outputs; MIT licensed — source (2026-08-26) verified
Ships as MCP server, CLI (npx octocode, 12 commands), a SKILL.md protocol, and a Pi agent extension — source (2026-08-26) verified

scorecard.json · transparency passport · registry · methodology

More: compare agents · best of · developer tools · incident registry

Verdict by Hlido Editor, our automated editorial system · Method: public-surface-tier-1+editorial-narrative-v2 · Methodology version 2026.05 · Next review due 2026-11-24

How this page was produced. The scores, claim verdicts and evidence come from automated hands-on testing of the product’s public surface. The written analysis is drafted by an AI system, and pages publish without a person reviewing each one. Hlido publishes this record and answers for it — tell us if anything here is wrong and we will correct it.

Embed this trust badge

Live, always-current independent score — free to embed on your site or README. No vendor pays for placement.

Markdown

[![Hlido trust score](https://hlido.eu/badge/bgauryy-octocode-mcp.svg)](https://hlido.eu/check/?agent=bgauryy-octocode-mcp)

HTML

<a href="https://hlido.eu/check/?agent=bgauryy-octocode-mcp"><img src="https://hlido.eu/badge/bgauryy-octocode-mcp.svg" alt="Hlido trust score"></a>