Add-ons that make your coding agent better

Memory, token savers, specs and skills that make Claude Code, Codex and other coding agents faster and cheaper.

For people who already use a coding agent every day and want it to remember more, waste fewer tokens and plan before it codes. Every pick works with Claude Code, and most work with Codex and other agents too.

10 picks in 6 sections · checked by an editor on Oct 30, 2026 · How we pick

01Memory1 pick

  1. Claude-Memthedotmack/claude-mem

    Pick this if your agent forgets what it learned in the last session.

    Extension / plugin · Apache-2.0 · active yesterday · ★ 106k

    99worth

02Save tokens3 picks

  1. RTKrtk-ai/rtk

    Pick this if most of your tokens go to output from git, tests and builds.

    CLI · Apache-2.0 · active yesterday · ★ 85k

    92worth
  2. Headroomheadroomlabs-ai/headroom

    Pick this to shrink large tool output before the model reads it.

    Library / SDK · Apache-2.0 · active yesterday · ★ 78k

    98worth
  3. CavemanJuliusBrussee/caveman

    Pick this if your agent writes long answers you do not read.

    CLI · Extension / plugin · Apache-2.0 · active 7 days ago · ★ 116k

    98worth

03Plan first2 picks

  1. Spec Kitgithub/spec-kit

    Pick this if you want the agent to write a spec before it writes code.

    CLI · MIT · active yesterday · ★ 144k

    92worth
  2. OpenSpecFission-AI/OpenSpec

    A lighter spec workflow that fits into an existing repo.

    CLI · MIT · active 23 hours ago · ★ 73k

    91worth

04Skills2 picks

  1. Superpowersobra/superpowers

    The most used skills library: test first, debug step by step, review before merge.

    Extension / plugin · MIT · active last week · ★ 309k

    88worth
  2. ECCaffaan-m/ECC

    Pick this for one package of skills, memory and guardrails instead of many.

    CLI · Extension / plugin · MIT · active 2 days ago · ★ 293k

    95worth

06Manage agents1 pick

  1. CC Switchfarion1231/cc-switch

    Pick this if you switch between Claude Code, Codex and Gemini CLI every day.

    Desktop · MIT · active yesterday · ★ 150k

    95worth

More in AI tools by job

Other groups

How we pick

An editor picks 6 to 12 projects for one job and writes when to choose each one. Every pick had a commit in the last 6 months. It is not archived. Its worth score is 60 or more. An editor breaks a rule only for a reason, and writes that reason in the pick's note. We check every list again each quarter. A pick goes when it breaks a rule or a clearly better project appears.