Commit
eeco
docs(readme): tighten tagline and design copy
Rewrite the tagline and intro, trim jargon ("terminal-integrated",
"no-AI-spend", "one cheap call"), and state the brief is built by a
local deterministic pass with no model call. Clarify single-binary
install (package manager or drop on PATH) and reword the design
bullets for plainer voice. modified README.md
@@ -4,7 +4,7 @@
<img src="assets/eeco_logo_light.png" alt="eeco" width="420">
</picture>
<h3>Single-binary, terminal-integrated workflow ecosystem + no-AI-spend knowledge layer for any AI assistant.</h3>
<h3>Single-binary workflow ecosystem and knowledge layer for any AI assistant.</h3>
<p>
<a href="https://github.com/ajhahnde/eeco/actions/workflows/ci.yml"><img src="https://github.com/ajhahnde/eeco/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
@@ -34,14 +34,15 @@
---
A single-binary, terminal-integrated, AI-assisted tool that gives any
coding project two things: a self-maintaining workflow ecosystem, and a
deterministic, no-AI-spend knowledge layer any AI assistant can plug
into. A control-center TUI, repeatable workflows that keep the project
hygienic and surface issues, a memory store with garbage collection, a
project brief and targeted Q&A that bring any assistant up to speed in
one cheap call, and an opt-in path to grow new workflows over time —
proposing larger changes for review rather than acting unilaterally.
eeco adds two things to any coding project, from the terminal: a
self-maintaining workflow ecosystem, and a deterministic, no-AI-spend
knowledge layer any AI assistant can read. It ships as a single static
binary. The workflow side is a control-center TUI, repeatable workflows
that keep the tracked tree hygienic and surface issues, and a memory
store with garbage collection. The knowledge side is a project brief and
targeted Q&A, both built by a local deterministic pass with no model
call. New workflows are added opt-in over time — proposed for review
rather than applied unilaterally.
<p align="center">
<picture>
@@ -52,8 +53,9 @@ proposing larger changes for review rather than acting unilaterally.
## Design
- **Single static binary, zero runtime dependencies.** Trivial to
install, embed, and ship; nothing to provision on the host.
- **Single static binary, zero runtime dependencies.** Install via a
package manager or drop the binary on `PATH`; nothing to provision on
the host.
- **Local-first and private.** eeco reads anywhere in a target repo but
writes only inside that repo's gitignored workspace. The sole
exception is `eeco init`, which may make one initial commit and push
@@ -62,8 +64,8 @@ proposing larger changes for review rather than acting unilaterally.
- **Pluggable AI, opt-in by default.** A provider interface with a
generic CLI provider wired in; every AI pass is gated by explicit
consent (`--ai` or `automation=auto`) and a per-invocation budget cap.
- **Not intrusive.** Nothing runs unless invoked. A single queue is the
only channel that asks for a decision.
- **Runs only when invoked.** A single queue is the only channel that
asks for a decision.
- **Reversible.** The two integrations that touch outside the workspace
(a local pre-commit hook, one entry in an AI CLI's user settings) are
opt-in and recorded in a ledger so they can be removed cleanly.
@@ -81,15 +83,14 @@ proposing larger changes for review rather than acting unilaterally.
pre-1.0 `cockpit-sync` workflow additionally keeps the generated AI
cockpit in step (see [`docs/COCKPIT.md`](docs/COCKPIT.md)).
- **Knowledge layer for AI assistants.** `eeco go` prints a
deterministic, no-AI-spend project brief (with `--json`, `--brief`,
`--write`, and `--copy` delivery axes); `eeco ask "<question>"`
deterministic project brief, built with no model call (with `--json`,
`--brief`, `--write`, and `--copy` delivery axes); `eeco ask "<question>"`
answers a targeted question with ranked `path:line` pointers;
`eeco add fact` / `eeco add task` let an assistant record what it
learns back into memory and the queue. The delivery channels are
brand-free — point Claude Code at `CLAUDE.md`, Gemini CLI at
`GEMINI.md`, Codex at `AGENTS.md`, Cursor at `.cursorrules`, or
paste `eeco go --copy` into any chat-only assistant. One cheap
call brings any assistant — not only the strongest — up to speed.
paste `eeco go --copy` into any chat-only assistant.
- **Memory store with garbage collection** (`eeco gc`). One fact per
file with flat frontmatter and a regenerated index.
- **The queue.** Every decision-bearing finding lands in one
@@ -105,8 +106,8 @@ proposing larger changes for review rather than acting unilaterally.
writes a user-global layer every project inherits, with per-repo
overrides — the git `--global` model. See
[`docs/USAGE.md`](docs/USAGE.md) §4a.
- **Diagnostics** (`eeco doctor`), **clean removal** (`eeco uninstall`), and **friction capture** (`eeco report-bug`) for every
step before, during, and after.
- **Diagnostics** (`eeco doctor`), **clean removal** (`eeco uninstall`),
and **friction capture** (`eeco report-bug`).
## Install