Dirac
Ranked 27 of 41 coding agents by recommenders
Open-source coding agent people recommend for token efficiency and fast refactoring on large codebases. Common caveat: it still feels rough, with bugs and missing integrations.
Open-source coding agent people recommend for token efficiency and fast refactoring on large codebases. Common caveat: it still feels rough, with bugs and missing integrations.
18 people gave an opinion
- recommend it
- 11
- criticize it
- 7
- approval share
- 61%
Recommended in 6 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: not recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 11 people recommend itNeeds 10 or more, met.
- 1.5 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 6 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Novel hash-anchored code editing
“Uses a novel approach to hash-anchoring that reduces the overhead of hash anchors to a minimum”
Comment from 9 Apr 2026 in “Show HN: I built Dirac, Hash Anchored AST native coding agent, costs -64.8 pct” “How come dirac's advance code editing technique is not being by Claude or Codex”
Comment from 31 May 2026 in “Cache hit rates of Inference are more meaningful than the headline costs” “The hashing and other optimizations in Direct seem kind of brilliant”
Comment from 29 Apr 2026 in “Claude system prompt bug wastes user money and bricks managed agents” Token-efficient, cheaper than alternatives
“The entire goal is to be token efficient (over 50% cheaper)”
Comment from 12 Apr 2026 in “Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage” “with the sole focus on token efficiency”
Comment from 24 Apr 2026 in “I Cancelled Claude: Token Issues, Declining Quality, and Poor Support” Productive on large refactoring tasks
“Dirac wins there significantly on those tasks because they are mostly refactoring related”
Comment from 12 Jul 2026 in “Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k” “dirac is more productive on large codebase refactoring than OpenCode”
Comment from 27 Apr 2026 in “Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview”
What people warn about
Each mark is one comment
Buggy, unfinished tooling
“The line-based anchors were slightly buggy though”
Comment from 6 Jun 2026 in “Show HN: Lowfat – pluggable CLI filter that saved 91.8% of my LLM tokens” “it's very cool but the tooling isn't there yet”
Comment from 27 Apr 2026 in “Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview” “quite OK but seems like full of bugs”
Comment from 3 Jul 2026 in “Jamesob's guide to running SOTA LLMs locally” Missing support for some logins and CLIs
“there isn't support for gemini-cli login”
Comment from 27 Apr 2026 in “Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview” “I keep trying to use dirac-cli with codex and it won't work”
Comment from 27 Apr 2026 in “Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview” Doubts about claimed token savings
“I couldn't even close to reproduce their benchmark results using the exact model and their harness.”
Comment from 11 Sep 2026 in “RTK reports token savings, but our cost benchmarks disagree” “My conclusion is that the efficiency dirac sees comes mainly from showing file skeleton by default”
Comment from 27 Apr 2026 in “Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview”
Other coding agents people recommend
The four most recommended other coding agents, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| GitHub Copilot | 571 | 993 | 37% | 12 | 1 yr46 owners | — |
| oh-my-pi | 104 | 28 | 79% | 9 | 2 mo5 owners | Approved |
| Grok Build CLI | 66 | 114 | 37% | 6 | — | — |
| DeepSeek Harness | 63 | 28 | 69% | 4 | 1 mo5 owners | Approved |
| DiracThis page | 11 | 7 | 61% | 6 | — | — |
What people said
24 of the 25 comments that mention it, newest first. Each links to the original comment.
“of course it will forget it in a long session and spit some too long text”
“I couldn't even close to reproduce their benchmark results using the exact model and their harness.”
“I recently added a 'build native tools on the fly' functionality to Dirac”
“Dirac wins there significantly on those tasks because they are mostly refactoring related”
“quite OK but seems like full of bugs”
“Dirac just launched (Woohoo!!)”
“The underlying code is clean but the smaller models often make boundary errors”
“The line-based anchors were slightly buggy though”
“How come dirac's advance code editing technique is not being by Claude or Codex”
“Glad to see it mentioned and even more glad that you found it useful.”
“The hashing and other optimizations in Direct seem kind of brilliant”
“the commit history stops at a gigantic "initial commit"”
“it's very cool but the tooling isn't there yet”
“there isn't support for gemini-cli login”
“This was the result of me stretching my capacity too thin and bulk renaming cline.bot to dirac.run”
“Always enabled, independent of telemetry opt-out. No way to disable without code changes.”
“My conclusion is that the efficiency dirac sees comes mainly from showing file skeleton by default”
“dirac is more productive on large codebase refactoring than OpenCode”
“Congratulations, great work.”
“I keep trying to use dirac-cli with codex and it won't work”
“Anyone working on this is anti-developer.”
“with the sole focus on token efficiency”
“The entire goal is to be token efficient (over 50% cheaper)”
“Uses a novel approach to hash-anchoring that reduces the overhead of hash anchors to a minimum”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.