kipimo 0.1.0__tar.gz

kipimo-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 AfriKaziOS maintainers
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

kipimo-0.1.0/PKG-INFO

@@ -0,0 +1,47 @@
+Metadata-Version: 2.4
+Name: kipimo
+Version: 0.1.0
+Summary: Swahili agent-task evaluation suite for the East Africa coordination stack — model-agnostic seed benchmark (46 tasks: server routing, term grounding, cascade routing).
+Author: AfriKaziOS maintainers
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/gabrielmahia/kipimo
+Project-URL: Repository, https://github.com/gabrielmahia/kipimo
+Project-URL: Issues, https://github.com/gabrielmahia/kipimo/issues
+Keywords: swahili,benchmark,evaluation,mcp,kenya,ai-agents
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# kipimo
+
+Over one hundred million people coordinate their lives in Swahili, yet no benchmark measures whether an AI agent can route their requests correctly — send money, check drought status, find a clinic, verify a worker's credentials. Agents targeting East Africa are evaluated on English tasks and deployed on faith.
+
+`kipimo` (Swahili: *a measure*) is a model-agnostic seed benchmark for exactly that gap: **46 tasks** across three types, with golds machine-derived from authoritative sources — the coordination-stack registry and the live `africa-coord-bus` routing table — never from memory.
+
+| Type | n | What it measures | Metric |
+|---|---|---|---|
+| `server_routing` | 25 | Swahili request → correct stack server (payments, tax, health, land, labour…) | exact |
+| `term_grounding` | 14 | Swahili domain term → English meaning | exact (case-insensitive) |
+| `cascade_routing` | 7 | Coordination event → which sectors must be notified | set F1 |
+
+## Use it (any model, no API keys)
+
+```bash
+pip install kipimo
+kipimo tasks > tasks.jsonl        # feed to your agent however you like
+kipimo template > preds.jsonl     # fill "prediction": [...] per id
+kipimo score preds.jsonl          # per-type + overall report
+```
+
+The harness never calls a model — you generate predictions with whatever system you're evaluating; kipimo only scores. Any lab can publish comparable numbers.
+
+## Honesty box
+- **v0.1 is a seed set.** 46 tasks establish the format and scoring; breadth comes from contributions.
+- Swahili phrasing is simple-register and **pending native-speaker review** — that is [issue #1](https://github.com/gabrielmahia/kipimo/issues), and corrections are the most valuable contribution possible.
+- Scores measure stack-routing competence, not general Swahili fluency.
+- Dataset: **CC BY 4.0** (usable by everyone, including commercial labs — that's the point). Harness: **MIT**.
+
+## IP & Collaboration
+
+MIT-licensed harness, CC BY 4.0 data. Feedback via GitHub Issues only — pull requests are not accepted; task corrections and additions via Issues are actively wanted. Full policy: [docs/architecture/IP_POLICY.md](docs/architecture/IP_POLICY.md). Security: see [SECURITY.md](SECURITY.md).

kipimo-0.1.0/README.md

@@ -0,0 +1,32 @@
+# kipimo
+
+Over one hundred million people coordinate their lives in Swahili, yet no benchmark measures whether an AI agent can route their requests correctly — send money, check drought status, find a clinic, verify a worker's credentials. Agents targeting East Africa are evaluated on English tasks and deployed on faith.
+
+`kipimo` (Swahili: *a measure*) is a model-agnostic seed benchmark for exactly that gap: **46 tasks** across three types, with golds machine-derived from authoritative sources — the coordination-stack registry and the live `africa-coord-bus` routing table — never from memory.
+
+| Type | n | What it measures | Metric |
+|---|---|---|---|
+| `server_routing` | 25 | Swahili request → correct stack server (payments, tax, health, land, labour…) | exact |
+| `term_grounding` | 14 | Swahili domain term → English meaning | exact (case-insensitive) |
+| `cascade_routing` | 7 | Coordination event → which sectors must be notified | set F1 |
+
+## Use it (any model, no API keys)
+
+```bash
+pip install kipimo
+kipimo tasks > tasks.jsonl        # feed to your agent however you like
+kipimo template > preds.jsonl     # fill "prediction": [...] per id
+kipimo score preds.jsonl          # per-type + overall report
+```
+
+The harness never calls a model — you generate predictions with whatever system you're evaluating; kipimo only scores. Any lab can publish comparable numbers.
+
+## Honesty box
+- **v0.1 is a seed set.** 46 tasks establish the format and scoring; breadth comes from contributions.
+- Swahili phrasing is simple-register and **pending native-speaker review** — that is [issue #1](https://github.com/gabrielmahia/kipimo/issues), and corrections are the most valuable contribution possible.
+- Scores measure stack-routing competence, not general Swahili fluency.
+- Dataset: **CC BY 4.0** (usable by everyone, including commercial labs — that's the point). Harness: **MIT**.
+
+## IP & Collaboration
+
+MIT-licensed harness, CC BY 4.0 data. Feedback via GitHub Issues only — pull requests are not accepted; task corrections and additions via Issues are actively wanted. Full policy: [docs/architecture/IP_POLICY.md](docs/architecture/IP_POLICY.md). Security: see [SECURITY.md](SECURITY.md).

kipimo-0.1.0/pyproject.toml

@@ -0,0 +1,33 @@
+[build-system]
+requires = ["setuptools>=77", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "kipimo"
+version = "0.1.0"
+description = "Swahili agent-task evaluation suite for the East Africa coordination stack — model-agnostic seed benchmark (46 tasks: server routing, term grounding, cascade routing)."
+readme = "README.md"
+requires-python = ">=3.10"
+license = "MIT"
+license-files = ["LICENSE"]
+authors = [{name = "AfriKaziOS maintainers"}]
+keywords = ["swahili", "benchmark", "evaluation", "mcp", "kenya", "ai-agents"]
+
+[project.urls]
+Homepage = "https://github.com/gabrielmahia/kipimo"
+Repository = "https://github.com/gabrielmahia/kipimo"
+Issues = "https://github.com/gabrielmahia/kipimo/issues"
+
+[project.scripts]
+kipimo = "kipimo.cli:_main"
+
+[tool.setuptools]
+package-dir = {"" = "src"}
+packages = ["kipimo"]
+
+[tool.setuptools.package-data]
+kipimo = ["data/*.jsonl"]
+
+[tool.ruff]
+target-version = "py310"
+line-length = 100

kipimo-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

kipimo-0.1.0/src/kipimo/__init__.py

@@ -0,0 +1,6 @@
+"""kipimo — Swahili agent-task evaluation for the East Africa coordination stack."""
+
+from .cli import __version__ as __version__
+from .cli import load_tasks as load_tasks
+from .cli import score_file as score_file
+from .cli import score_one as score_one

kipimo-0.1.0/src/kipimo/cli.py

@@ -0,0 +1,115 @@
+"""kipimo — Swahili agent-task evaluation for the East Africa coordination stack.
+
+Model-agnostic by design: kipimo emits tasks and scores prediction files. It
+never calls a model API, so any lab, student, or vendor can evaluate any agent
+against the same gold set. Golds are machine-derived from authoritative
+sources (the stack registry; the africa-coord-bus routing table), never from
+memory.
+
+Usage:
+    kipimo tasks                 # emit the task set (JSONL) to stdout
+    kipimo template              # emit an empty predictions file to fill in
+    kipimo score preds.jsonl     # score predictions against gold
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from collections import defaultdict
+from importlib import resources
+
+__version__ = "0.1.0"
+
+DISCLAIMER = ("kipimo v0.1 is a SEED benchmark (46 tasks). Swahili phrasing is "
+              "simple-register and pending native-speaker review (issue #1). "
+              "Scores indicate stack-routing competence, not general Swahili "
+              "fluency. Do not use as a sole deployment gate.")
+
+
+def load_tasks() -> list[dict]:
+    text = resources.files("kipimo").joinpath("data/kipimo_v0.1.jsonl").read_text("utf-8")
+    return [json.loads(line) for line in text.splitlines() if line.strip()]
+
+
+def _norm(s: str) -> str:
+    return " ".join(str(s).lower().split())
+
+
+def score_one(task: dict, pred: list[str]) -> float:
+    gold = [_norm(g) for g in task["gold"]]
+    p = [_norm(x) for x in (pred or [])]
+    if task["metric"] in ("exact", "exact_ci"):
+        return 1.0 if p and p[0] in gold else 0.0
+    if task["metric"] == "set_f1":
+        gs, ps = set(gold), set(p)
+        if not ps:
+            return 0.0
+        tp = len(gs & ps)
+        prec, rec = tp / len(ps), tp / len(gs)
+        return 0.0 if tp == 0 else 2 * prec * rec / (prec + rec)
+    raise ValueError(f"unknown metric {task['metric']}")
+
+
+def score_file(path: str) -> dict:
+    preds = {}
+    with open(path, encoding="utf-8") as f:
+        for line in f:
+            if line.strip():
+                row = json.loads(line)
+                preds[row["id"]] = row.get("prediction", [])
+    tasks = load_tasks()
+    by_type: dict[str, list[float]] = defaultdict(list)
+    missing = []
+    for t in tasks:
+        if t["id"] not in preds:
+            missing.append(t["id"])
+            by_type[t["type"]].append(0.0)
+        else:
+            by_type[t["type"]].append(score_one(t, preds[t["id"]]))
+    report = {k: round(sum(v) / len(v), 4) for k, v in by_type.items()}
+    allv = [x for v in by_type.values() for x in v]
+    report["overall"] = round(sum(allv) / len(allv), 4)
+    report["n_tasks"] = len(tasks)
+    report["n_missing"] = len(missing)
+    return report
+
+
+def main(argv: list[str] | None = None) -> int:
+    p = argparse.ArgumentParser(prog="kipimo", description=__doc__.split("\n")[0],
+                                epilog=DISCLAIMER)
+    sub = p.add_subparsers(dest="cmd", required=True)
+    sub.add_parser("tasks", help="emit task set JSONL to stdout")
+    sub.add_parser("template", help="emit empty predictions JSONL to stdout")
+    sp = sub.add_parser("score", help="score a predictions file")
+    sp.add_argument("predictions", help="JSONL with {id, prediction:[...]} rows")
+    args = p.parse_args(argv)
+
+    if args.cmd == "tasks":
+        for t in load_tasks():
+            print(json.dumps(t, ensure_ascii=False))
+    elif args.cmd == "template":
+        for t in load_tasks():
+            print(json.dumps({"id": t["id"], "prediction": []}))
+    else:
+        rep = score_file(args.predictions)
+        print(json.dumps(rep, indent=2))
+        print(f"\n{DISCLAIMER}", file=sys.stderr)
+        if rep["n_missing"]:
+            print(f"Note: {rep['n_missing']} task(s) had no prediction and scored 0. "
+                  f"Run `kipimo template` for the full id list.", file=sys.stderr)
+    return 0
+
+
+def _main() -> int:
+    try:
+        return main()
+    except BrokenPipeError:
+        import os
+        os.dup2(os.open(os.devnull, os.O_WRONLY), sys.stdout.fileno())
+        return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

kipimo-0.1.0/src/kipimo/data/kipimo_v0.1.jsonl

@@ -0,0 +1,46 @@
+{"id": "rt-01", "type": "server_routing", "lang": "sw", "input": "Nataka kutuma pesa kwa mama yangu kupitia simu.", "options": null, "gold": ["mpesa-mcp"], "metric": "exact"}
+{"id": "rt-02", "type": "server_routing", "lang": "sw", "input": "Ninahitaji kujua deni langu la ushuru KRA.", "options": null, "gold": ["kra-mcp"], "metric": "exact"}
+{"id": "rt-03", "type": "server_routing", "lang": "sw", "input": "Nataka mkopo mdogo kwa biashara yangu ya mboga.", "options": null, "gold": ["mkopo-mcp"], "metric": "exact"}
+{"id": "rt-04", "type": "server_routing", "lang": "sw", "input": "Je, kuna bima ya mazao dhidi ya ukame?", "options": null, "gold": ["bima-mcp"], "metric": "exact"}
+{"id": "rt-05", "type": "server_routing", "lang": "sw", "input": "Bei ya mahindi sokoni Nakuru ni ngapi leo?", "options": null, "gold": ["soko-mcp"], "metric": "exact"}
+{"id": "rt-06", "type": "server_routing", "lang": "sw", "input": "Nataka kutuma pesa kutoka Marekani hadi Kenya kwa gharama nafuu.", "options": null, "gold": ["remit-mcp"], "metric": "exact"}
+{"id": "rt-07", "type": "server_routing", "lang": "sw", "input": "Ni wapi hospitali iliyo karibu inayotoa huduma za upasuaji?", "options": null, "gold": ["kenya-health-mcp"], "metric": "exact"}
+{"id": "rt-08", "type": "server_routing", "lang": "sw", "input": "Mtoto wangu ana homa na kikohozi, nifanye nini?", "options": null, "gold": ["afya-mcp"], "metric": "exact"}
+{"id": "rt-09", "type": "server_routing", "lang": "sw", "input": "Ninahisi huzuni sana na msongo wa mawazo, nipate msaada wapi?", "options": null, "gold": ["afya-ya-akili-mcp"], "metric": "exact"}
+{"id": "rt-10", "type": "server_routing", "lang": "sw", "input": "Nipande mahindi lini msimu huu Uasin Gishu?", "options": null, "gold": ["kilimo-mcp"], "metric": "exact"}
+{"id": "rt-11", "type": "server_routing", "lang": "sw", "input": "Hali ya ukame kaunti ya Baringo ikoje sasa?", "options": null, "gold": ["wapimaji-mcp"], "metric": "exact"}
+{"id": "rt-12", "type": "server_routing", "lang": "sw", "input": "Nahitaji kibali cha mazingira kwa kiwanda changu kidogo.", "options": null, "gold": ["mazingira-mcp"], "metric": "exact"}
+{"id": "rt-13", "type": "server_routing", "lang": "sw", "input": "Huduma za kaunti ya Machakos ni zipi na bajeti yake?", "options": null, "gold": ["county-mcp"], "metric": "exact"}
+{"id": "rt-14", "type": "server_routing", "lang": "sw", "input": "Nifanye nini kupata cheti cha kuzaliwa? Fomu gani inahitajika?", "options": null, "gold": ["fomu-mcp"], "metric": "exact"}
+{"id": "rt-15", "type": "server_routing", "lang": "sw", "input": "Mwajiri wangu amenifukuza bila notisi. Haki zangu ni zipi?", "options": null, "gold": ["haki-ya-kazi-mcp"], "metric": "exact"}
+{"id": "rt-16", "type": "server_routing", "lang": "sw", "input": "Natafuta kazi ya udereva Nairobi.", "options": null, "gold": ["kazi-mcp"], "metric": "exact"}
+{"id": "rt-17", "type": "server_routing", "lang": "sw", "input": "Shule za sekondari Kiambu na matokeo ya KCSE.", "options": null, "gold": ["elimu-mcp"], "metric": "exact"}
+{"id": "rt-18", "type": "server_routing", "lang": "sw", "input": "Tafsiri sentensi hii kwa Kiingereza tafadhali.", "options": null, "gold": ["tafsiri-mcp"], "metric": "exact"}
+{"id": "rt-19", "type": "server_routing", "lang": "sw", "input": "Nataka kuhakiki ujuzi na vyeti vya mfanyakazi ninayemwajiri.", "options": null, "gold": ["sifa-mcp"], "metric": "exact"}
+{"id": "rt-20", "type": "server_routing", "lang": "sw", "input": "Nyaraka za shamba langu Kiambu — nithibitishe umiliki.", "options": null, "gold": ["ardhi-mcp"], "metric": "exact"}
+{"id": "rt-21", "type": "server_routing", "lang": "sw", "input": "Nyumba ya kupanga Ruaka bei gani kwa mwezi?", "options": null, "gold": ["nyumba-mcp"], "metric": "exact"}
+{"id": "rt-22", "type": "server_routing", "lang": "sw", "input": "Njia za matatu kutoka CBD hadi Rongai?", "options": null, "gold": ["usafiri-mcp"], "metric": "exact"}
+{"id": "rt-23", "type": "server_routing", "lang": "sw", "input": "Niko diaspora — nasimamia vipi mali yangu Kenya nikiwa nje?", "options": null, "gold": ["diaspora-mcp"], "metric": "exact"}
+{"id": "rt-24", "type": "server_routing", "lang": "sw", "input": "Tunataka kuanzisha chama cha kuweka akiba mtaani.", "options": null, "gold": ["jumuia-mcp"], "metric": "exact"}
+{"id": "rt-25", "type": "server_routing", "lang": "sw", "input": "Umeme haujafika kijiji chetu — kuna mpango gani wa nishati?", "options": null, "gold": ["nishati-mcp"], "metric": "exact"}
+{"id": "tm-01", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'mkopo' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["credit/loan"], "metric": "exact_ci"}
+{"id": "tm-02", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'bima' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["insurance"], "metric": "exact_ci"}
+{"id": "tm-03", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'faida' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["profit/returns"], "metric": "exact_ci"}
+{"id": "tm-04", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'soko' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["market"], "metric": "exact_ci"}
+{"id": "tm-05", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'afya' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["health"], "metric": "exact_ci"}
+{"id": "tm-06", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'afya ya akili' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["mental health"], "metric": "exact_ci"}
+{"id": "tm-07", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'kilimo' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["agriculture"], "metric": "exact_ci"}
+{"id": "tm-08", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'wapimaji' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["water"], "metric": "exact_ci"}
+{"id": "tm-09", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'mazingira' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["environment"], "metric": "exact_ci"}
+{"id": "tm-10", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'fomu' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["forms"], "metric": "exact_ci"}
+{"id": "tm-11", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'haki ya kazi' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["labour rights"], "metric": "exact_ci"}
+{"id": "tm-12", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'kazi' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["work"], "metric": "exact_ci"}
+{"id": "tm-13", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'habari' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["news"], "metric": "exact_ci"}
+{"id": "tm-14", "type": "term_grounding", "lang": "sw", "input": "Neno la Kiswahili 'elimu' linamaanisha nini kwa Kiingereza (neno moja au mawili)?", "gold": ["education"], "metric": "exact_ci"}
+{"id": "cas-01", "type": "cascade_routing", "lang": "sw", "input": "Ukame umefikia hatua ya tahadhari Baringo. Taarifa hii inapaswa kufika kwenye sekta zipi?", "event": {"domain": "water", "event_type": "drought_alert", "severity": "alert"}, "gold": ["agriculture", "finance", "health"], "metric": "set_f1"}
+{"id": "cas-02", "type": "cascade_routing", "lang": "sw", "input": "Ukame ni hatari sana Turkana. Sekta zipi zinapaswa kuarifiwa?", "event": {"domain": "water", "event_type": "drought_alert", "severity": "critical"}, "gold": ["agriculture", "finance", "health"], "metric": "set_f1"}
+{"id": "cas-03", "type": "cascade_routing", "lang": "sw", "input": "Mlipuko wa kipindupindu Kisumu. Taarifa ipelekwe kwenye sekta zipi?", "event": {"domain": "health", "event_type": "disease_outbreak", "severity": "alert"}, "gold": ["civic", "procurement"], "metric": "set_f1"}
+{"id": "cas-04", "type": "cascade_routing", "lang": "sw", "input": "Onyo la mafuriko bonde la Tana. Sekta zipi zihusishwe?", "event": {"domain": "water", "event_type": "flood_alert", "severity": "warning"}, "gold": ["civic", "health"], "metric": "set_f1"}
+{"id": "cas-05", "type": "cascade_routing", "lang": "sw", "input": "Mafuriko makubwa Tana River. Nani apokee taarifa?", "event": {"domain": "water", "event_type": "flood_alert", "severity": "critical"}, "gold": ["civic", "health"], "metric": "set_f1"}
+{"id": "cas-06", "type": "cascade_routing", "lang": "sw", "input": "Mlipuko mkubwa wa ugonjwa. Sekta zipi zinahusika?", "event": {"domain": "health", "event_type": "disease_outbreak", "severity": "critical"}, "gold": ["civic", "procurement"], "metric": "set_f1"}
+{"id": "cas-07", "type": "cascade_routing", "lang": "sw", "input": "Dalili za ukame zimeanza Kajiado. Wapi taarifa ipelekwe?", "event": {"domain": "water", "event_type": "drought_alert", "severity": "warning"}, "gold": ["agriculture", "finance"], "metric": "set_f1"}

kipimo-0.1.0/src/kipimo.egg-info/PKG-INFO

@@ -0,0 +1,47 @@
+Metadata-Version: 2.4
+Name: kipimo
+Version: 0.1.0
+Summary: Swahili agent-task evaluation suite for the East Africa coordination stack — model-agnostic seed benchmark (46 tasks: server routing, term grounding, cascade routing).
+Author: AfriKaziOS maintainers
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/gabrielmahia/kipimo
+Project-URL: Repository, https://github.com/gabrielmahia/kipimo
+Project-URL: Issues, https://github.com/gabrielmahia/kipimo/issues
+Keywords: swahili,benchmark,evaluation,mcp,kenya,ai-agents
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# kipimo
+
+Over one hundred million people coordinate their lives in Swahili, yet no benchmark measures whether an AI agent can route their requests correctly — send money, check drought status, find a clinic, verify a worker's credentials. Agents targeting East Africa are evaluated on English tasks and deployed on faith.
+
+`kipimo` (Swahili: *a measure*) is a model-agnostic seed benchmark for exactly that gap: **46 tasks** across three types, with golds machine-derived from authoritative sources — the coordination-stack registry and the live `africa-coord-bus` routing table — never from memory.
+
+| Type | n | What it measures | Metric |
+|---|---|---|---|
+| `server_routing` | 25 | Swahili request → correct stack server (payments, tax, health, land, labour…) | exact |
+| `term_grounding` | 14 | Swahili domain term → English meaning | exact (case-insensitive) |
+| `cascade_routing` | 7 | Coordination event → which sectors must be notified | set F1 |
+
+## Use it (any model, no API keys)
+
+```bash
+pip install kipimo
+kipimo tasks > tasks.jsonl        # feed to your agent however you like
+kipimo template > preds.jsonl     # fill "prediction": [...] per id
+kipimo score preds.jsonl          # per-type + overall report
+```
+
+The harness never calls a model — you generate predictions with whatever system you're evaluating; kipimo only scores. Any lab can publish comparable numbers.
+
+## Honesty box
+- **v0.1 is a seed set.** 46 tasks establish the format and scoring; breadth comes from contributions.
+- Swahili phrasing is simple-register and **pending native-speaker review** — that is [issue #1](https://github.com/gabrielmahia/kipimo/issues), and corrections are the most valuable contribution possible.
+- Scores measure stack-routing competence, not general Swahili fluency.
+- Dataset: **CC BY 4.0** (usable by everyone, including commercial labs — that's the point). Harness: **MIT**.
+
+## IP & Collaboration
+
+MIT-licensed harness, CC BY 4.0 data. Feedback via GitHub Issues only — pull requests are not accepted; task corrections and additions via Issues are actively wanted. Full policy: [docs/architecture/IP_POLICY.md](docs/architecture/IP_POLICY.md). Security: see [SECURITY.md](SECURITY.md).

kipimo-0.1.0/src/kipimo.egg-info/SOURCES.txt

@@ -0,0 +1,13 @@
+LICENSE
+README.md
+pyproject.toml
+src/kipimo/__init__.py
+src/kipimo/cli.py
+src/kipimo.egg-info/PKG-INFO
+src/kipimo.egg-info/SOURCES.txt
+src/kipimo.egg-info/dependency_links.txt
+src/kipimo.egg-info/entry_points.txt
+src/kipimo.egg-info/top_level.txt
+src/kipimo/data/kipimo_v0.1.jsonl
+tests/test_kipimo.py
+tests/test_smoke.py

kipimo-0.1.0/src/kipimo.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

kipimo-0.1.0/src/kipimo.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+kipimo = kipimo.cli:_main

kipimo-0.1.0/src/kipimo.egg-info/top_level.txt

@@ -0,0 +1 @@
+kipimo

kipimo-0.1.0/tests/test_kipimo.py

@@ -0,0 +1,47 @@
+"""Integrity + scoring tests for the kipimo seed benchmark."""
+import json
+import subprocess
+import sys
+
+from kipimo import load_tasks, score_one
+
+
+def test_dataset_integrity():
+    tasks = load_tasks()
+    assert len(tasks) == 46
+    ids = [t["id"] for t in tasks]
+    assert len(ids) == len(set(ids)), "duplicate ids"
+    for t in tasks:
+        assert t["type"] in ("server_routing", "term_grounding", "cascade_routing")
+        assert t["gold"], t["id"]
+        assert t["metric"] in ("exact", "exact_ci", "set_f1")
+        assert t["input"].strip()
+
+
+def test_exact_scoring():
+    t = {"gold": ["mpesa-mcp"], "metric": "exact"}
+    assert score_one(t, ["mpesa-mcp"]) == 1.0
+    assert score_one(t, ["MPESA-MCP"]) == 1.0  # normalized
+    assert score_one(t, ["kra-mcp"]) == 0.0
+    assert score_one(t, []) == 0.0
+
+
+def test_set_f1_scoring():
+    t = {"gold": ["health", "finance"], "metric": "set_f1"}
+    assert score_one(t, ["health", "finance"]) == 1.0
+    assert 0 < score_one(t, ["health"]) < 1.0
+    assert score_one(t, ["water"]) == 0.0
+
+
+def test_cli_end_to_end(tmp_path):
+    r = subprocess.run([sys.executable, "-m", "kipimo.cli", "template"],
+                       capture_output=True, text=True)
+    assert r.returncode == 0
+    preds = tmp_path / "p.jsonl"
+    # perfect predictions = gold copied in
+    lines = [json.dumps({"id": t["id"], "prediction": t["gold"]}) for t in load_tasks()]
+    preds.write_text("\n".join(lines))
+    r = subprocess.run([sys.executable, "-m", "kipimo.cli", "score", str(preds)],
+                       capture_output=True, text=True)
+    rep = json.loads(r.stdout)
+    assert rep["overall"] == 1.0 and rep["n_missing"] == 0

kipimo-0.1.0/tests/test_smoke.py

@@ -0,0 +1,12 @@
+import ast
+import pathlib
+
+
+def test_all_sources_parse():
+    root = pathlib.Path(__file__).parent.parent / "src"
+    for f in root.rglob("*.py"):
+        ast.parse(f.read_text(encoding="utf-8"))
+
+
+def test_package_importable():
+    import kipimo  # noqa: F401