Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,7 @@ Hive uses an **AI Codebase Maturity Model** (ACMM) with six levels that control
| L3 | Quality-Gated (Measured) | Quality agent opens issues and PRs gated by the literal `hold` level-gate label. Others remain advisory. |
| L4 | Security-Aware (Adaptive) | All agents file issues. Quality, sec-check, and CI open PRs gated by `hold`. |
| L5 | Semi-Autonomous (Semi-Automated) | All agents open PRs gated by `hold`. Humans batch-review and approve. |
| L6 | Fully Autonomous | Agents open PRs and auto-merge when checks pass. Non-outreach PRs have no level hold; outreach PRs remain held. |
| L6 | Autonomous | Agents open PRs and auto-merge when checks pass. Non-outreach PRs have no level hold; outreach PRs remain held. |

See [src/docs/acmm-policy-matrix.md](src/docs/acmm-policy-matrix.md) for the full matrix and [src/docs/labels-and-control-signals.md](src/docs/labels-and-control-signals.md) for label behavior.

Expand Down
2 changes: 1 addition & 1 deletion bin/hive-setup.sh
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@
# L1: (local tooling only β€” no hive needed)
# L2: supervisor scanner reviewer (analysis + issue filing, NO PRs)
# L3: supervisor scanner reviewer architect (opens PRs)
# L4: supervisor scanner reviewer architect outreach sec-check (full autonomy)
# L4: supervisor scanner reviewer architect outreach sec-check (autonomous)
# L5+: supervisor scanner reviewer architect outreach sec-check strategist quality

set -euo pipefail
Expand Down
1 change: 1 addition & 0 deletions changelog.d/changed-l6-autonomous-wording.md
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
- ACMM L6 is now labelled **Autonomous** rather than "Fully Autonomous" everywhere operators see it β€” the dashboard header pill (`L6 Β· Autonomous`), the maturity-level dialog, the hub's level picker and ACMM pages, the `level-6` pack name, advisor criteria text, the tool-approval desk rationale, and the docs. L6 still parks items for human decisions, honours `hold`, cannot grant secrets, org membership, or workflow pushes, and keeps an owner in the loop, so "fully" overstated what the level does. The ladder is otherwise unchanged (L1 Inception β†’ L5 Semi-Autonomous β†’ L6 Autonomous); the numeric `acmm_level: 6` config value and the pack filename `level-6.yaml` are untouched, so no configuration changes are needed.
2 changes: 1 addition & 1 deletion dashboard/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -3444,7 +3444,7 @@ <h4 style="font-size:0.82rem;color:#f0f6fc;margin:14px 0 6px 0;">Agents</h4>
const titles = {
observe: 'Data collection only β€” governor unchanged',
suggest: 'Experiments require your approval',
evolve: 'Fully autonomous experimentation',
evolve: 'Autonomous experimentation',
};
html += `<button class="nous-mode-btn${active}" title="${titles[m]}" onclick="nousSetMode('${m}')">${m.charAt(0).toUpperCase() + m.slice(1)}</button>`;
}
Expand Down
4 changes: 2 additions & 2 deletions docs/presentations/hive-hub-and-spoke.html
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@
<div class="slide center active" id="s1">
<h1>Hive: Hub &amp; Spoke</h1>
<p class="accent" style="font-size:1.8rem;font-weight:600;margin-bottom:24px">AI Agent Orchestration for Open Source</p>
<p style="margin-bottom:4px">From brainstorming to full autonomy and everything in-between.</p>
<p style="margin-bottom:4px">From brainstorming to autonomous operation and everything in-between.</p>
<p style="margin-bottom:32px">A fully customizable fleet of AI agents covering every level of project maintenance.</p>
<p style="color:#8b949e">hive.hivecommons.dev</p>
</div>
Expand Down Expand Up @@ -97,7 +97,7 @@ <h1>Agents &amp; the Governor</h1>
<!-- Slide 4: ACMM Levels & the Proxy -->
<div class="slide" id="s4">
<h1>ACMM Levels &amp; the Proxy</h1>
<p>The <span class="accent">AI Codebase Maturity Model</span> - 6 levels from brainstorming to fully autonomous.</p>
<p>The <span class="accent">AI Codebase Maturity Model</span> - 6 levels from brainstorming to autonomous.</p>
<p style="margin-bottom:16px">The <span class="accent">ACMM Proxy</span> sits between agents and GitHub, enforcing level constraints and preventing rogue behavior.</p>
<table>
<tr><th>Level</th><th>Name</th><th>Agents</th><th>Allowed Actions</th></tr>
Expand Down
2 changes: 1 addition & 1 deletion examples/acmm/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,6 @@ These Markdown fragments document the operating constraints for each AI Codebase
| [l3.md](l3.md) | L3 Quality-Gated | Measured quality-lane behavior plus the reviewer on the hold-gated PR queue. |
| [l4.md](l4.md) | L4 Security-Aware | Issues plus selected hold-gated PR behavior; reviewer on the hold-gated PR queue. |
| [l5.md](l5.md) | L5 Semi-Autonomous | Hold-gated PR behavior for trusted lanes. |
| [l6.md](l6.md) | L6 Fully Autonomous | Auto-merge-on-green constraints. |
| [l6.md](l6.md) | L6 Autonomous | Auto-merge-on-green constraints. |

See [src/docs/acmm-policy-matrix.md](../../src/docs/acmm-policy-matrix.md) for the authoritative lane-by-lane policy matrix used by the built-in packs.
6 changes: 3 additions & 3 deletions examples/acmm/l6.md
Original file line number Diff line number Diff line change
@@ -1,8 +1,8 @@
# ACMM Level 6 β€” Fully Autonomous
# ACMM Level 6 β€” Autonomous

**This file OVERRIDES any conflicting instruction in your agent policy.**

You are operating at **ACMM Level 6**. The project grants full autonomous operation.
You are operating at **ACMM Level 6**. The project grants autonomous operation.

## Permissions

Expand All @@ -24,4 +24,4 @@ At L6, you own the full lifecycle. The operator trusts you to:

## Budget Awareness

Full autonomy does not mean unlimited spending. Respect the governor's token budget and mode signals. If budget is low, focus on high-impact work and defer low-priority items.
Autonomy does not mean unlimited spending. Respect the governor's token budget and mode signals. If budget is low, focus on high-impact work and defer low-priority items.
2 changes: 1 addition & 1 deletion examples/kubestellar/auto-qa-skill.md
Original file line number Diff line number Diff line change
Expand Up @@ -41,7 +41,7 @@ Process items in this priority order:

---

### 2. For Each Item β€” Full Autonomous Workflow
### 2. For Each Item β€” Autonomous Workflow

#### 2a. Close Stalled Copilot PR (if one exists for this issue)

Expand Down
2 changes: 1 addition & 1 deletion src/cmd/hive/approvalhook.go
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ import (
type hookApprovalAdapter struct {
inbox *toolapprove.Inbox
// acmmLevel is the hive's level, recorded on the queued item so the panel
// can flag an approval parked on a full-autonomy hive as a probable policy
// can flag an approval parked on an L6 (Autonomous) hive as a probable policy
// bug (throughput-contract requirement 4) exactly as it does for the
// sweep-produced ones.
acmmLevel int
Expand Down
4 changes: 2 additions & 2 deletions src/cmd/hive/outreach_hold_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -16,8 +16,8 @@ func TestShouldHoldAgentPR(t *testing.T) {
ownsRepo func(agent, repo string) bool
want bool
}{
{name: "cross-repo outreach is held at full autonomy", agent: "outreach", repo: "someone-else/awesome-list", level: 6, ownsRepo: ownsRepo, want: true},
{name: "same-repo outreach is not held at full autonomy", agent: "outreach", repo: "hive", level: 6, ownsRepo: ownsRepo, want: false},
{name: "cross-repo outreach is held at L6", agent: "outreach", repo: "someone-else/awesome-list", level: 6, ownsRepo: ownsRepo, want: true},
{name: "same-repo outreach is not held at L6", agent: "outreach", repo: "hive", level: 6, ownsRepo: ownsRepo, want: false},
{name: "same-repo outreach accepts an org-qualified target", agent: "outreach", repo: "hivecommons/hive", level: 6, ownsRepo: ownsRepo, want: false},
{name: "same-repo outreach still follows the level gate", agent: "outreach", repo: "console", level: 4, ownsRepo: ownsRepo, want: true},
{name: "outreach identity is normalized", agent: " Outreach ", repo: "other/repo", level: 6, ownsRepo: ownsRepo, want: true},
Expand Down
8 changes: 4 additions & 4 deletions src/crux/toolapprove/decide.go
Original file line number Diff line number Diff line change
Expand Up @@ -165,11 +165,11 @@ func Decide(ctx context.Context, req ToolRequest, acmmLevel int, agentId AgentId
Agent: agentName,
}

default: // L6 Full Autonomy
// L6 Full: side-effectful tools route through security scan, auto-approve on green
default: // L6 Autonomous
// L6 Autonomous: side-effectful tools route through security scan, auto-approve on green
return Verdict{
Decision: DecisionSecurityScan,
Rationale: "ACMM L6 (full autonomy): side-effectful tool routes through pre-execution security scan; auto-approves on green",
Rationale: "ACMM L6 (Autonomous): side-effectful tool routes through pre-execution security scan; auto-approves on green",
Tool: toolName,
ACMMLevel: acmmLevel,
Agent: agentName,
Expand Down Expand Up @@ -213,7 +213,7 @@ func Resolve(ctx context.Context, req ToolRequest, acmmLevel int, agentId AgentI
switch {
case acmmLevel >= 6:
v.Decision = DecisionAutoApprove
v.Rationale = "ACMM L6 full autonomy: auto-approved on green security scan"
v.Rationale = "ACMM L6 (Autonomous): auto-approved on green security scan"
case acmmLevel == 5:
v.Decision = DecisionAutoApprove
v.Rationale = "ACMM L5 hold-gated: auto-approved on green security scan"
Expand Down
2 changes: 1 addition & 1 deletion src/crux/toolapprove/desk.go
Original file line number Diff line number Diff line change
Expand Up @@ -115,7 +115,7 @@ type RuleMatch struct {
//
// L0-L2 operator-approve β€” nothing side-effectful resolves without a human
// L3-L5 security-scan β€” may reach auto-approve only via a green scan
// L6+ auto-approve β€” full autonomy; the desk is an audit record, not a gate
// L6+ auto-approve β€” autonomous; the desk is an audit record, not a gate
//
// An unknown/negative level clamps to the most restrictive lane, so a
// mis-parsed or absent level can never widen authority.
Expand Down
8 changes: 4 additions & 4 deletions src/crux/toolapprove/inbox.go
Original file line number Diff line number Diff line change
Expand Up @@ -452,12 +452,12 @@ func (i *Inbox) ResolveMany(ids []string, approved bool, operator, rationale str
return out
}

// PendingAtFullAutonomy returns the pending items queued at an ACMM level that
// claims full autonomy. Per requirement 4 of the throughput contract, these are
// PendingAtAutonomousLevel returns the pending items queued at an ACMM level that
// is Autonomous (L6). Per requirement 4 of the throughput contract, these are
// surfaced as PROBABLE POLICY BUGS rather than silently waiting: an inbox that
// accumulates on a hive whose level says "full autonomy" is a misconfiguration
// accumulates on a hive whose level says "Autonomous" is a misconfiguration
// signal. The API marks them and the dashboard renders them distinctly.
func (i *Inbox) PendingAtFullAutonomy() []PendingItem {
func (i *Inbox) PendingAtAutonomousLevel() []PendingItem {
var out []PendingItem
for _, p := range i.List() {
if p.ACMMLevel >= 6 {
Expand Down
12 changes: 6 additions & 6 deletions src/crux/toolapprove/inbox_l6_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -146,19 +146,19 @@ func TestL6RuleAutoApproveNeverQueues(t *testing.T) {
}
}

// TestPendingAtFullAutonomyIsFlagged pins requirement 4: an item queued on a
// TestPendingAtAutonomousLevelIsFlagged pins requirement 4: an item queued on a
// hive at L6 is surfaced as a probable policy bug rather than silently waiting.
func TestPendingAtFullAutonomyIsFlagged(t *testing.T) {
func TestPendingAtAutonomousLevelIsFlagged(t *testing.T) {
inbox := NewInbox(filepath.Join(t.TempDir(), "inbox.json"))

if _, err := inbox.Enqueue(sideEffectfulRequest(), Verdict{
Decision: DecisionOperatorApprove, ACMMLevel: 6,
}); err != nil {
t.Fatalf("Enqueue: %v", err)
}
bugs := inbox.PendingAtFullAutonomy()
bugs := inbox.PendingAtAutonomousLevel()
if len(bugs) != 1 {
t.Fatalf("PendingAtFullAutonomy returned %d items, want 1 β€” an inbox accumulating at L6 "+
t.Fatalf("PendingAtAutonomousLevel returned %d items, want 1 β€” an inbox accumulating at L6 "+
"is a misconfiguration signal that must be surfaced, not queued politely", len(bugs))
}

Expand All @@ -172,8 +172,8 @@ func TestPendingAtFullAutonomyIsFlagged(t *testing.T) {
if _, err := inbox.Enqueue(other, Verdict{Decision: DecisionOperatorApprove, ACMMLevel: 3}); err != nil {
t.Fatalf("Enqueue: %v", err)
}
if got := len(inbox.PendingAtFullAutonomy()); got != 1 {
t.Errorf("an L3 pending item was flagged as a policy bug: PendingAtFullAutonomy=%d, want 1", got)
if got := len(inbox.PendingAtAutonomousLevel()); got != 1 {
t.Errorf("an L3 pending item was flagged as a policy bug: PendingAtAutonomousLevel=%d, want 1", got)
}
}

Expand Down
4 changes: 2 additions & 2 deletions src/docs/acmm-advisor.md
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ The advisor evaluates a fixed set of six signals (`Signals` struct, `src/pkg/acm
| `HoldCount` | Count of open PRs still carrying a `hold` label awaiting human review. |
| `HasQualityAgent` | Whether an active quality/coverage agent is present in the config. |

Levels run L1 (Assisted) through L6 (Fully Autonomous), matching the level pack files and the dashboard's 1–6 bound (`src/pkg/acmmadvisor/acmmadvisor.go:34-42`). The advisor never proposes a target more than one level above the current level β€” advancement is always incremental (`levelStep = 1`, same file).
Levels run L1 (Assisted) through L6 (Autonomous), matching the level pack files and the dashboard's 1–6 bound (`src/pkg/acmmadvisor/acmmadvisor.go:34-42`). The advisor never proposes a target more than one level above the current level β€” advancement is always incremental (`levelStep = 1`, same file).

### Where the live signals come from

Expand All @@ -42,7 +42,7 @@ The advisor computes the checklist for advancing to the level **one above** the
| L3 (Measured) | required | β‰₯ 40% | β‰₯ 3 | β€” | β€” | β€” |
| L4 (Adaptive) | required | β‰₯ 60% | β‰₯ 5 | β‰₯ 70% | β€” | β€” |
| L5 (Semi-Automated) | required | β‰₯ 75% | β‰₯ 8 | β‰₯ 85% | ≀ 25 | ≀ 15 |
| L6 (Fully Autonomous) | required | β‰₯ 90% | β‰₯ 12 | β‰₯ 95% | ≀ 10 | ≀ 5 |
| L6 (Autonomous) | required | β‰₯ 90% | β‰₯ 12 | β‰₯ 95% | ≀ 10 | ≀ 5 |

The L2 pack includes a COMMENT-only `reviewer` as the review-first entry point, and the reviewer ships in every pack from L2 up (L3/L4 review the `hold`-gated queue with no merge power; L5/L6 per [#8023](https://github.com/hivecommons/hive/issues/8023)); the advisor criteria above are unchanged by it. Decision taken by hive: review is part of L2 proper (COMMENT-only); no L2.5 level β€” a new level would ripple through level enums, advisor and dashboard; keeping L2 triage+review preserves the review-first entry point the vote asked for.

Expand Down
2 changes: 1 addition & 1 deletion src/docs/acmm-policy-matrix.md
Original file line number Diff line number Diff line change
Expand Up @@ -101,7 +101,7 @@ Agents open issues AND pull requests. Agent PRs get literal `hold` from the leve
| operations (paused) | holdgated | `operations-holdgated.md` |
| brainstorm | advisory | `brainstorm-advisory.md` |

### L6 β€” Fully Autonomous (15 agents)
### L6 β€” Autonomous (15 agents)

Existing autonomous lanes can open issues, create PRs, and auto-merge on green CI. Auto-merge is off below L6; switching to L6 turns it on for every active repository, after which owners can toggle repositories individually. L6 PRs do not get the level hold, except outreach PRs into a repository the hive does not manage, which are still held for human review; outreach PRs into the hive's own repos auto-merge like any other lane's. Contributor PRs can merge through the contributor lane when the author has an operator-trusted tier or the conservative `auto_merge.contributor_tier.auto_promote_after_merged` clean-history ratchet admits them; reviewer verdict, CI, diff fences, soak and branch protection still apply. Hive never needs App Actions write for `action_required` workflow runs: when trusted same-repo bot/App/registry-trusted clanker PRs hit the workflow-approval banner, Hive posts `/ok-to-test` once per head and the repository's self-contained `hive-ok-to-test` workflow approves those runs with its `GITHUB_TOKEN`; if the workflow is absent or a `sentinel-alert` is present, Hive surfaces `workflow-approval-required:<ctx>` / stage 085 for a human decision. Outreach handles community engagement. Reviewer stays advisory unless the self-authored review gate is enforcing: `review.require_approval: true` (or `governor.review_gate.mode: enforce`) makes the App self-authored sweep require same-head aggregate approval, while `requires_human`/incomplete verdicts can be cleared by a trusted human's normal GitHub Approve review on that exact head. For Hive-authored PRs, `review.self_authored_decision_policy: resolve` (the default) lets the reviewer resolve every decision question itself except those in a named risk class (permissions/credentials, security, sensitive paths, data loss/irreversible, merge/trust policy, external side effects, licensing/legal, unverifiable) or marked `hive-decision-risk: high`; set `escalate` to park every decision question at 085. Swarm `changes_requested`/`reject` still requires a fix or direct human merge. Use actual holds or GitHub branch rules for a universal merge stop. Adjudicator is the reviewer lane: it repairs, de-escalates, or recommends closing (and may close) escalated `needs-human` hive PRs, and never merges. Dedupe posts the same daily consolidation checklist as at L5 and stays recommendation-only. Telemetry and operations remain paused and use `ISSUES_AND_PRS`, so they never merge their own PRs.

Expand Down
2 changes: 1 addition & 1 deletion src/docs/agent-configuration.md
Original file line number Diff line number Diff line change
Expand Up @@ -775,7 +775,7 @@ You don't have to design a roster. Hive ships six **ACMM packs** (`level-1.yaml`
| L3 | Quality-Gated (Measured) | quality opens issues and test PRs gated by literal `hold`; the rest stay advisory |
| L4 | Security-Aware (Adaptive) | scanner/guide file issues; quality, ci-maintainer, and sec-check can open PRs gated by literal `hold` |
| L5 | Semi-Autonomous (Semi-Automated) | issues **and** PRs; PRs are gated by literal `hold`; humans batch-approve |
| L6 | Fully Autonomous | auto-merge on green CI; no level hold, except cross-repo outreach PRs (target repo not managed by this hive) remain held |
| L6 | Autonomous | auto-merge on green CI; no level hold, except cross-repo outreach PRs (target repo not managed by this hive) remain held |

Applying a level **reconciles the whole roster**, not just the diff: missing agents are created (as overlay files in `/data/agent-configs/`), existing agents are merged β€” pack values fill blanks, but your explicit `backend:`, `model:`, and `enabled: false` always win β€” and the level's `kick_template`, `mode` and `on_demand` are updated so the agent's *policy* matches the level (an `on_demand` you toggled yourself in the agent's settings dialog is operator-owned and left alone; leaving on-demand starts the agent, entering it stops it). A failed agent doesn't abort the rest; the level is only recorded as cleanly applied when every agent reconciled.

Expand Down
4 changes: 2 additions & 2 deletions src/docs/architecture.md
Original file line number Diff line number Diff line change
Expand Up @@ -293,7 +293,7 @@ flowchart LR
L3["L3 Quality-Gated<br/>quality opens PRs"]
L4["L4 Security-Aware<br/>more agents file/PR"]
L5["L5 Semi-Autonomous<br/>all PRs, hold gated"]
L6["L6 Fully Autonomous<br/>auto-merge switches on"]
L6["L6 Autonomous<br/>auto-merge switches on"]
L1 --> L2 --> L3 --> L4 --> L5 --> L6
```

Expand All @@ -304,7 +304,7 @@ flowchart LR
| L3 | Quality-Gated (Measured) | `quality` opens PRs gated by literal `hold` about testing gaps, coverage, and CI health; others advisory. This measurement foundation is what earns automation at higher levels |
| L4 | Security-Aware (Adaptive) | All agents file issues; quality, ci-maintainer, and sec-check can open PRs gated by literal `hold` |
| L5 | Semi-Autonomous (Semi-Automated) | All agents open PRs β€” every PR carries literal `hold` for human review |
| L6 | Fully Autonomous | Agents open PRs and **auto-merge on green CI**; switching to L6 enables auto-merge for every active repo, owners may toggle repos afterward, no level hold, except cross-repo outreach PRs (target repo not managed by this hive) remain held |
| L6 | Autonomous | Agents open PRs and **auto-merge on green CI**; switching to L6 enables auto-merge for every active repo, owners may toggle repos afterward, no level hold, except cross-repo outreach PRs (target repo not managed by this hive) remain held |

`supervisor` is always advisory. The full matrix is in
[`acmm-policy-matrix.md`](acmm-policy-matrix.md).
Expand Down
2 changes: 1 addition & 1 deletion src/docs/cncf-reference-architecture.md
Original file line number Diff line number Diff line change
Expand Up @@ -132,7 +132,7 @@ Two ideas define the architecture:
judgment calls (read code, reason about a fix, write a PR).

2. **Autonomy is a dial, enforced in depth.** An **AI-native Capability Maturity
Model (ACMM)** with six levels (advisory β†’ fully autonomous) is the single
Model (ACMM)** with six levels (advisory β†’ autonomous) is the single
control a human turns. Each level maps deterministically to a per-agent
permission mode, and that mode is enforced at three independent layers: CLI
tool denial, least-privilege scoped GitHub App tokens, and a network-level
Expand Down
Loading
Loading