Why contribute to Engram
Shared agent memory is an open problem. Individual agent memory is
solved — every major IDE and framework handles session context.
What happens when two agents on the same team develop
contradictory beliefs? What happens when a session ends and the
next agent starts from zero, re-discovering decisions that were
already made and settled? That's what Engram is built to fix.
Strong contributors receive paid contract offers ($125–185/hour,
weekly pay, 1099). See ./HIRING.md.
Where to start
Three tiers depending on how much design work you want to do
before writing code.
Tier 1: Claim a pre-scoped task (easiest path)
These are already specified. Comment on the issue to claim it,
build it, open a PR.
- Add a quickstart guide for an unsupported IDE — we have Claude
Code, Cursor, Windsurf, VS Code Copilot, and Zed. Any AI coding
environment without a guide in docs/quickstart/ is fair game.
- Improve a vague error message — identify one, trace it to its
source, make it actionable.
- Add test coverage for an edge case you noticed while reading the
code.
Browse the https://github.com/Agentscreator/Engram/labels/good%20first%20issue label for what's currently open.
Tier 2: Pick a small feature or roadmap item
These are defined in the README roadmap and the issue tracker.
Still pre-scoped, slightly bigger surface:
- Conflict detection quality — the detection pipeline is 5-tier
and working, but precision on semantic contradictions has room to
improve. NLP background is useful here.
- Fact lineage UI — the engine tracks full fact provenance; the
dashboard doesn't expose it yet.
- Performance under load — indexing and query planning for
workspaces with tens of thousands of facts.
Browse https://github.com/Agentscreator/Engram/labels/good%20first%20issue or https://github.com/Agentscreator/Engram/labels/P2 for
what's open.
Tier 3: Propose a domain integration or vertical use case (highest
leverage)
If you have real depth in a domain — legal, medical, fintech,
education, game dev, security research — and want a signature
contribution with your name on it, propose a vertical integration
or example that shows Engram working in your context.
This path requires a written proposal and design discussion before
code. Higher bar than Tiers 1 and 2, higher payoff: the kind of
contribution you can walk a hiring manager through, and one that
can live in the README under your name.
Bigger ideas
Open an issue with a short design note first. We prefer to align
on scope before you write code. If you've built something real on
top of Engram and want to contribute a sanitized version, open a
discussion — we'll find the right home for it.
How we review
Read ./CONTRIBUTING.md before opening a PR. Every PR needs:
passing tests (pytest tests/ -x), Ruff clean (uv run ruff check),
and a short PR body covering what changed and why. Small, focused
PRs move fast. Large refactors or core API changes go through
design discussion first.
What we won't merge
- AI-generated filler: reformatting PRs, content without real
substance
- New runtime dependencies without a compelling reason — we hold
to three
- Duplicates: search open issues and PRs before starting
- Changes to the core API surface without a prior discussion
- PRs without a description
Credit
Every merged contributor is listed in release notes with handle
and PR link. Sustained contributions can lead to co-maintaining a
module and to paid contract work. See ./HIRING.md for how that
path works.
Getting help
Use https://github.com/Agentscreator/Engram/discussions for
questions. Comment on the issue if you're unsure what to pick up.
If you're stuck mid-PR, tag @Agentscreator in a comment — blockers
don't age well.
Why contribute to Engram
Shared agent memory is an open problem. Individual agent memory is
solved — every major IDE and framework handles session context.
What happens when two agents on the same team develop
contradictory beliefs? What happens when a session ends and the
next agent starts from zero, re-discovering decisions that were
already made and settled? That's what Engram is built to fix.
Strong contributors receive paid contract offers ($125–185/hour,
weekly pay, 1099). See ./HIRING.md.
Where to start
Three tiers depending on how much design work you want to do
before writing code.
Tier 1: Claim a pre-scoped task (easiest path)
These are already specified. Comment on the issue to claim it,
build it, open a PR.
Code, Cursor, Windsurf, VS Code Copilot, and Zed. Any AI coding
environment without a guide in docs/quickstart/ is fair game.
source, make it actionable.
code.
Browse the https://github.com/Agentscreator/Engram/labels/good%20first%20issue label for what's currently open.
Tier 2: Pick a small feature or roadmap item
These are defined in the README roadmap and the issue tracker.
Still pre-scoped, slightly bigger surface:
and working, but precision on semantic contradictions has room to
improve. NLP background is useful here.
dashboard doesn't expose it yet.
workspaces with tens of thousands of facts.
Browse https://github.com/Agentscreator/Engram/labels/good%20first%20issue or https://github.com/Agentscreator/Engram/labels/P2 for
what's open.
Tier 3: Propose a domain integration or vertical use case (highest
leverage)
If you have real depth in a domain — legal, medical, fintech,
education, game dev, security research — and want a signature
contribution with your name on it, propose a vertical integration
or example that shows Engram working in your context.
This path requires a written proposal and design discussion before
code. Higher bar than Tiers 1 and 2, higher payoff: the kind of
contribution you can walk a hiring manager through, and one that
can live in the README under your name.
Bigger ideas
Open an issue with a short design note first. We prefer to align
on scope before you write code. If you've built something real on
top of Engram and want to contribute a sanitized version, open a
discussion — we'll find the right home for it.
How we review
Read ./CONTRIBUTING.md before opening a PR. Every PR needs:
passing tests (pytest tests/ -x), Ruff clean (uv run ruff check),
and a short PR body covering what changed and why. Small, focused
PRs move fast. Large refactors or core API changes go through
design discussion first.
What we won't merge
substance
to three
Credit
Every merged contributor is listed in release notes with handle
and PR link. Sustained contributions can lead to co-maintaining a
module and to paid contract work. See ./HIRING.md for how that
path works.
Getting help
Use https://github.com/Agentscreator/Engram/discussions for
questions. Comment on the issue if you're unsure what to pick up.
If you're stuck mid-PR, tag @Agentscreator in a comment — blockers
don't age well.