Skip to content

XY-1118: [ELF AMK P6] Add Work Continuity benchmark suite and hard-fail gates #251

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
yvette-carlisle merged 1 commit into from
Jun 27, 2026
Merged

XY-1118: [ELF AMK P6] Add Work Continuity benchmark suite and hard-fail gates #251

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
52 changes: 52 additions & 0 deletions Makefile.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,6 +82,9 @@
# | real-world-memory-summary | composite | |
# | real-world-memory-summary-json | command | |
# | real-world-memory-summary-report | command | |
# | real-world-memory-work-continuity | composite | |
# | real-world-memory-work-continuity-json | command | |
# | real-world-memory-work-continuity-report | command | |

[tasks.baseline-backfill-100k-docker]
workspace = false
Expand Down Expand Up @@ -1225,6 +1228,55 @@ args = [
"tmp/real-world-memory/scheduled/report.md",
]

[tasks.real-world-memory-work-continuity]
workspace = false
dependencies = [
"real-world-memory-work-continuity-report",
]

[tasks.real-world-memory-work-continuity-json]
workspace = false
command = "cargo"
args = [
"run",
"-p",
"elf-eval",
"--bin",
"real_world_job_benchmark",
"--",
"run",
"--fixtures",
"apps/elf-eval/fixtures/real_world_memory/work_continuity",
"--out",
"tmp/real-world-memory/work-continuity/report.json",
"--run-id",
"real-world-memory-work-continuity",
"--adapter-id",
"fixture_work_continuity",
"--adapter-name",
"ELF Work Continuity fixture",
]

[tasks.real-world-memory-work-continuity-report]
workspace = false
dependencies = [
"real-world-memory-work-continuity-json",
]
command = "cargo"
args = [
"run",
"-p",
"elf-eval",
"--bin",
"real_world_job_benchmark",
"--",
"publish",
"--report",
"tmp/real-world-memory/work-continuity/report.json",
"--out",
"tmp/real-world-memory/work-continuity/report.md",
]

[tasks.real-world-memory-service-native-dreaming]
workspace = false
command = "bash"
Expand Down
151 changes: 151 additions & 0 deletions apps/elf-eval/fixtures/real_world_memory/work_continuity/decision_rationale_readback.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,151 @@
{
"schema": "elf.real_world_job/v1",
"job_id": "work-continuity-decision-rationale-001",
"suite": "work_continuity",
"title": "Decision rationale survives Work Journal readback",
"corpus": {
"corpus_id": "real-world-memory-work-continuity-2026-06-27",
"profile": "synthetic",
"items": [
{
"evidence_id": "wj-decision-rationale-source",
"kind": "decision_rationale",
"text": "Decision rationale: keep Work Journal entries source-adjacent because current facts require Memory Authority promotion or accepted Dreaming Review.",
"source_ref": {
"schema": "source_ref/v1",
"resolver": "real_world_job_fixture/v1",
"ref": {
"fixture": "decision_rationale_readback",
"evidence_id": "wj-decision-rationale-source"
},
"locator": {
"quote": "source-adjacent"
}
},
"created_at": "2026-06-27T00:02:00Z"
}
],
"adapter_response": {
"adapter_id": "fixture_work_continuity",
"answer": {
"content": "The rationale is to keep Work Journal entries source-adjacent unless they are promoted through Memory Authority or accepted Dreaming Review.",
"claims": [
{
"claim_id": "decision_rationale_recall",
"text": "The decision rationale for Work Journal is source-adjacent capture until explicit promotion.",
"evidence_ids": ["wj-decision-rationale-source"],
"confidence": "high"
}
],
"evidence_ids": ["wj-decision-rationale-source"],
"work_journal_readbacks": [
{
"readback_id": "wj-readback-decision-rationale",
"contract_schema": "elf.work_journal/v1",
"generated_at": "2026-06-27T00:03:00Z",
"session_id": "work-continuity-session",
"tenant_id": "fixture-tenant",
"project_id": "elf",
"agent_id": "xy-1118-fixture-agent",
"read_profile": "private_plus_project",
"items": [
{
"entry_id": "wj-decision-entry",
"family": "decision_rationale",
"title": "Work Journal authority boundary",
"body": "Keep entries source-adjacent until accepted promotion.",
"source_refs": ["wj-decision-rationale-source"]
}
],
"where_stopped": {
"reset_resume_entry_ids": [],
"decision_rationale_evidence_ids": ["wj-decision-rationale-source"],
"current_explicit_next_step_ids": [],
"labeled_inferred_next_step_ids": [],
"handoff_source_refs": ["wj-decision-rationale-source"],
"journal_only_authority_claims": []
},
"promotion_boundary": {
"journal_entry_authority": "source_adjacent_only",
"memory_promotion_required": true,
"accepted_refs": []
},
"janitor_candidates": []
}
],
"latency_ms": 1.0,
"cost": {
"currency": "USD",
"amount": 0.0,
"input_tokens": 0,
"output_tokens": 0
}
}
}
},
"prompt": {
"role": "user",
"content": "Why is the Work Journal not current fact authority?",
"job_mode": "work_continuity",
"constraints": ["cite_evidence", "recall_decision_rationale"]
},
"expected_answer": {
"must_include": [
{
"claim_id": "decision_rationale_recall",
"text": "The decision rationale for Work Journal is source-adjacent capture until explicit promotion."
}
],
"must_not_include": ["Work Journal alone is the Memory Authority"],
"evidence_links": {
"decision_rationale_recall": ["wj-decision-rationale-source"]
},
"answer_type": "work_journal_readback",
"accepted_alternates": [],
"requires_caveat": false,
"requires_refusal": false
},
"required_evidence": [
{
"evidence_id": "wj-decision-rationale-source",
"claim_id": "decision_rationale_recall",
"requirement": "cite",
"quote": "source-adjacent"
}
],
"work_continuity": {
"required_decision_rationale_evidence_ids": ["wj-decision-rationale-source"]
},
"scoring_rubric": {
"dimensions": {
"answer_correctness": {
"weight": 0.35,
"max_points": 1.0,
"criteria": "Recalls the stored rationale."
},
"evidence_grounding": {
"weight": 0.3,
"max_points": 1.0,
"criteria": "Cites the rationale evidence."
},
"trap_avoidance": {
"weight": 0.2,
"max_points": 1.0,
"criteria": "Does not promote journal-only content."
},
"lifecycle_behavior": {
"weight": 0.15,
"max_points": 1.0,
"criteria": "Preserves promotion boundary."
}
},
"pass_threshold": 0.85,
"hard_fail_rules": ["journal-only authority claim"]
},
"allowed_uncertainty": {
"can_answer_unknown": false,
"acceptable_phrases": [],
"fallback_action": "state_blocker"
},
"tags": ["synthetic", "work_continuity", "decision_rationale"]
}
160 changes: 160 additions & 0 deletions apps/elf-eval/fixtures/real_world_memory/work_continuity/explicit_next_step_precision.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,160 @@
{
"schema": "elf.real_world_job/v1",
"job_id": "work-continuity-explicit-next-step-001",
"suite": "work_continuity",
"title": "Explicit next steps are returned as instructions",
"corpus": {
"corpus_id": "real-world-memory-work-continuity-2026-06-27",
"profile": "synthetic",
"items": [
{
"evidence_id": "wj-explicit-step-source",
"kind": "next_step",
"text": "Explicit next step: run cargo make real-world-memory-work-continuity after adding the fixtures.",
"source_ref": {
"schema": "source_ref/v1",
"resolver": "real_world_job_fixture/v1",
"ref": {
"fixture": "explicit_next_step_precision",
"evidence_id": "wj-explicit-step-source"
},
"locator": {
"quote": "Explicit next step"
}
},
"created_at": "2026-06-27T00:06:00Z"
}
],
"adapter_response": {
"adapter_id": "fixture_work_continuity",
"answer": {
"content": "The explicit next step is to run cargo make real-world-memory-work-continuity after adding the fixtures.",
"claims": [
{
"claim_id": "explicit_next_step",
"text": "The explicit next step is run cargo make real-world-memory-work-continuity.",
"evidence_ids": ["wj-explicit-step-source"],
"confidence": "high"
}
],
"evidence_ids": ["wj-explicit-step-source"],
"work_journal_readbacks": [
{
"readback_id": "wj-readback-explicit-next-step",
"contract_schema": "elf.work_journal/v1",
"generated_at": "2026-06-27T00:07:00Z",
"session_id": "work-continuity-session",
"tenant_id": "fixture-tenant",
"project_id": "elf",
"agent_id": "xy-1118-fixture-agent",
"read_profile": "private_plus_project",
"items": [
{
"entry_id": "wj-explicit-step-entry",
"family": "next_step",
"title": "Run Work Continuity slice",
"body": "The next step was explicitly recorded by the operator.",
"source_refs": ["wj-explicit-step-source"],
"explicit_next_steps": [
{
"step_id": "wj-explicit-run-slice",
"text": "Run cargo make real-world-memory-work-continuity.",
"label": "explicit",
"instruction": true,
"evidence_refs": ["wj-explicit-step-source"]
}
]
}
],
"where_stopped": {
"reset_resume_entry_ids": [],
"decision_rationale_evidence_ids": [],
"current_explicit_next_step_ids": ["wj-explicit-run-slice"],
"labeled_inferred_next_step_ids": [],
"handoff_source_refs": ["wj-explicit-step-source"],
"journal_only_authority_claims": []
},
"promotion_boundary": {
"journal_entry_authority": "source_adjacent_only",
"memory_promotion_required": true,
"accepted_refs": []
},
"janitor_candidates": []
}
],
"latency_ms": 1.0,
"cost": {
"currency": "USD",
"amount": 0.0,
"input_tokens": 0,
"output_tokens": 0
}
}
}
},
"prompt": {
"role": "user",
"content": "What is the explicit next step?",
"job_mode": "work_continuity",
"constraints": ["cite_evidence", "label_explicit_next_steps"]
},
"expected_answer": {
"must_include": [
{
"claim_id": "explicit_next_step",
"text": "The explicit next step is run cargo make real-world-memory-work-continuity."
}
],
"must_not_include": ["This step was inferred"],
"evidence_links": {
"explicit_next_step": ["wj-explicit-step-source"]
},
"answer_type": "work_journal_readback",
"accepted_alternates": [],
"requires_caveat": false,
"requires_refusal": false
},
"required_evidence": [
{
"evidence_id": "wj-explicit-step-source",
"claim_id": "explicit_next_step",
"requirement": "cite",
"quote": "Explicit next step"
}
],
"work_continuity": {
"required_explicit_next_step_ids": ["wj-explicit-run-slice"]
},
"scoring_rubric": {
"dimensions": {
"answer_correctness": {
"weight": 0.35,
"max_points": 1.0,
"criteria": "Returns only correct explicit next steps."
},
"evidence_grounding": {
"weight": 0.3,
"max_points": 1.0,
"criteria": "Cites the explicit step evidence."
},
"trap_avoidance": {
"weight": 0.2,
"max_points": 1.0,
"criteria": "Does not mislabel explicit steps."
},
"lifecycle_behavior": {
"weight": 0.15,
"max_points": 1.0,
"criteria": "Keeps source ref attached."
}
},
"pass_threshold": 0.85,
"hard_fail_rules": ["missing explicit next step"]
},
"allowed_uncertainty": {
"can_answer_unknown": false,
"acceptable_phrases": [],
"fallback_action": "state_blocker"
},
"tags": ["synthetic", "work_continuity", "explicit_next_step"]
}
Loading