Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Routing Protocol Analysis · automateyournetwork | DecimalAI

automateyournetwork/pyats-routing

SkillScore77Maintained

CCIE-level routing protocol analysis - OSPF, BGP, EIGRP, IS-IS, static routes, RIB/FIB verification, redistribution audit, and convergence validation. Use when analyzing routing tables, debugging OSPF neighbors, checking BGP peering, verifying route redistribution, or validating convergence after changes.

DevOpsby automateyournetwork

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

DevOps & SRE pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/automateyournetwork-pyats-routing/SKILL.md

decimalai skills pull automateyournetwork-pyats-routing --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/automateyournetwork-pyats-routing/SKILL.md https://app.decimal.ai/s/automateyournetwork-pyats-routing/SKILL.md

Raw file: /s/automateyournetwork-pyats-routing/SKILL.md ↗ · machine-readable summary at /s/automateyournetwork-pyats-routing.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 20d ago

—+152% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-04	✗→✓	▲ Improved	59%	0%
case-06	✗→✓	▲ Improved	278%	0%
case-08	✗→✓	▲ Improved	260%	0%
case-10	✗→✓	▲ Improved	179%	0%
case-13	✗→✓	▲ Improved	41%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

automateyournetwork / netclaw

workspace/skills/pyats-routing/SKILL.md

View on GitHub ↗

License Apache-2.0Imported 8/3/2026Version v1

Skill contents

SKILL.md

Routing Protocol Analysis

Routing Table Analysis

Always start here. The routing table is the source of truth for forwarding decisions.

Full Routing Table

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route"}'

Per-Protocol Routes

bash
# OSPF routes only
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route ospf"}'

# BGP routes only
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route bgp"}'

# Connected and static
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route connected"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route static"}'

VRF Routes

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route vrf MGMT"}'

Analysis checklist:

Route source codes: C (connected), S (static), O (OSPF), O IA (OSPF inter-area), O E1/E2 (OSPF external), B (BGP), D (EIGRP), D EX (EIGRP external)
Administrative distance and metric for each route
Next-hop reachability — is the next-hop IP actually reachable?
Recursive lookups — routes pointing to next-hops resolved through other routes
Equal-cost multipath (ECMP) — multiple next-hops for the same prefix
Default route presence and source
Unexpected route absence — if a prefix should be there but is not

OSPF Deep Dive

OSPF Process Overview

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf"}'

Check: Router ID, areas configured, SPF run count (high = instability), reference bandwidth, stub/NSSA config, authentication.

OSPF Neighbors

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf neighbor"}'

Neighbor state analysis:

FULL — Healthy, full adjacency
FULL/DR or FULL/BDR — Normal on broadcast/NBMA networks
FULL/DROTHER — Normal on broadcast/NBMA (non-DR/BDR)
2WAY/DROTHER — Normal, two DROTHERs don't form full adjacency
INIT — One-way communication only → check MTU, ACLs, authentication
EXSTART/EXCHANGE — Stuck in database exchange → MTU mismatch (most common), OSPF area mismatch, authentication
LOADING — Stuck loading LSAs → database corruption, memory issues
DOWN — No hello packets → interface down, ACL blocking, hello/dead timer mismatch

Common OSPF problems and their symptoms: | Symptom | Likely Cause | |---------|-------------| | Stuck in EXSTART | MTU mismatch between neighbors | | Stuck in INIT | Hello reaching neighbor but not returning (ACL, asymmetric routing) | | Neighbor flapping | Unstable link, hello/dead timer too aggressive, CPU too high to process hellos | | Missing routes | Area type mismatch (stub vs non-stub), missing redistribute or default-information originate | | Suboptimal routing | Cost misconfiguration, missing auto-cost reference-bandwidth |

OSPF Interfaces

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf interface"}'

Check per interface: Area assignment, network type (broadcast/point-to-point/NBMA), cost, hello/dead timers, DR/BDR election, authentication type, passive status.

OSPF Database (LSDB)

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf database"}'

LSA types to understand:

Type 1 (Router LSA) — Every router generates one per area
Type 2 (Network LSA) — DR generates on broadcast/NBMA segments
Type 3 (Summary LSA) — ABR advertises between areas
Type 4 (ASBR Summary) — ABR advertises ASBR reachability
Type 5 (External LSA) — ASBR redistributes external routes
Type 7 (NSSA External) — External routes in NSSA areas (converted to Type 5 at ABR)

Red flags in LSDB:

Rapidly incrementing sequence numbers → LSA flooding loop
Type 5 LSAs in stub area → misconfiguration
Missing Type 3 LSAs for expected inter-area prefixes → ABR filtering or area mismatch
Duplicate Router IDs → same RID on two routers

BGP Deep Dive

BGP Summary

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip bgp summary"}'

Check: Local AS, router ID, table version, total paths/best paths, per-neighbor state.

Neighbor state column meaning:

Number (e.g., 5) → Established, number = received prefixes
Idle → Not configured correctly, or administratively shut
Idle (Admin) → neighbor shutdown configured
Active → TCP connection failing → check reachability, TTL, ACLs
Connect → TCP SYN sent, no response
OpenSent → TCP connected, waiting for OPEN reply
OpenConfirm → OPEN received, waiting for KEEPALIVE
Established → Healthy

BGP Neighbors Detail

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip bgp neighbors"}'

Deep analysis per neighbor:

BGP state, uptime, last reset reason
Messages sent/received (OPEN, UPDATE, KEEPALIVE, NOTIFICATION)
Hold time and keepalive interval
Address families negotiated
Route-map/filter-list/prefix-list applied (inbound/outbound)
Next-hop-self, soft-reconfiguration, route-reflector-client status
Notification messages — decode the error code/subcode

Common BGP problems: | Symptom | Likely Cause | |---------|-------------| | Stuck in Active | TCP connection failing — check ACL, reachability, update-source, ebgp-multihop | | Neighbor flapping | Unstable link, route-map causing route churn, max-prefix exceeded | | 0 prefixes received | No network or redistribute on remote side, outbound filter on remote, address-family not activated | | Routes not in RIB | Next-hop unreachable (next-hop-self missing for iBGP), route filtered by policy | | Suboptimal path | Weight/local-pref/AS-path/MED not set correctly |

BGP Table

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip bgp"}'

BGP best path selection order (memorize this):

Highest Weight (Cisco proprietary, local to router)
Highest Local Preference (iBGP, default 100)
Locally originated (network/redistribute/aggregate)
Shortest AS-path
Lowest Origin (IGP < EGP < Incomplete)
Lowest MED (from same neighbor AS only)
eBGP over iBGP
Lowest IGP metric to next-hop
Oldest route (for eBGP)
Lowest router ID
Lowest neighbor IP

EIGRP Analysis

EIGRP Neighbors

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip eigrp neighbors"}'

Check: Hold timer (resetting to max means hellos received), uptime, SRTT, RTO, Q count (should be 0 — non-zero means retransmission queue backed up).

EIGRP Topology

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip eigrp topology"}'

DUAL states:

Passive (P) → Route is stable, normal
Active (A) → Route in active query state → DUAL is searching for a feasible successor → if stuck, SIA (Stuck In Active) will kill the neighbor

EIGRP metric components (classic): Bandwidth, Delay, Reliability, Load, MTU (only BW and delay used by default).

EIGRP metric components (wide/named mode): Throughput, Latency, Reliability, Load, MTU, Extended attributes.

Redistribution Audit

When multiple routing protocols are in use, check redistribution points:

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip protocols"}'

Redistribution checklist:

Is redistribution mutual (two-way)? Risk of routing loops
Are route-maps/prefix-lists applied to redistribution? They should be
Are metrics set correctly on redistribution? (seed metric for EIGRP, metric-type for OSPF)
Is there a risk of suboptimal routing through the redistribution boundary?
Administrative distance tuning — are AD values set to prefer the right protocol?

Route Filtering Verification

bash
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show route-map"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip prefix-list"}'

Convergence Validation

After any routing change, verify convergence:

Check all expected neighbor adjacencies are FULL/Established
Verify the routing table has all expected prefixes
Ping across the network from the device to validate data plane
Check logs for any protocol events during the change
Compare route counts before and after

bash
# Verify route count
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "${PYATS_PYTHON:-python3} -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route summary"}'

Want to see if automateyournetwork/pyats-routing would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/16/2026, 12:26 UTC · 22 cases

Measured on 18 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+27%

77% vs 50%

Agent turns

1.0 vs 1.0

Tokens

+152%

4,222.44 vs 1,673.39

fail → pass · 7pass → pass · 10fail → fail · 4pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-18 A BGP peer session between R1 and R2 reaches Established state, but R1 reports receiving 0 prefixes from R2. What configuration items on R2 or R1 account for this?	pass→pass	14,181	11,985	-15%	1	1	0%	2,567	5,387	+110%	0	0	—
case-19 While inspecting the OSPF LSDB on R1 via show ip ospf database, what specific sequence number pattern signals an active LSA flooding loop in the area?	pass→pass	10,773	3,962	-63%	1	1	0%	1,985	3,682	+85%	0	0	—
case-01 Can you analyze the OSPF setup on device R1? I need a detailed diagnostic report covering neighbor adjacencies, interface configurations, and database status to figure out why neighbor relationships might be failing to reach full state.	fail→fail	9,062	27,217	+200%	1	1	0%	1,540	3,307	+115%	0	0	—
case-02 Please perform a BGP health check on device R2. Provide a summary breakdown of peer session states, neighbor details, and prefix tables to identify any down peers or issues with route propagation.	fail→fail	10,940	5,428	-50%	1	1	0%	1,883	3,252	+73%	0	0	—
case-03 We are auditing network convergence on device Core-GW1. Please examine the overall IP routing table, analyze EIGRP neighbor topologies, and review protocol redistribution settings to detect missing prefixes or potential routing loop risks.	fail→fail	16,369	4,676	-71%	1	1	0%	2,756	3,274	+19%	0	0	—
case-04 Device R3's OSPF neighbor relationship with R4 is staying in INIT state rather than progressing to FULL. How should we investigate and diagnose this issue on R3 using testbed execution?	fail→pass	15,664	8,772	-44%	1	1	0%	2,946	4,684	+59%	0	0	—
case-05 A network router has two iBGP paths learned for prefix 10.50.0.0/16. Path A has Local Preference 200 and AS-path length 3. Path B has Weight 500, Local Preference 100, and AS-path length 5. Which path will Cisco BGP select as best path, and why?	pass→pass	5,409	6,404	+18%	1	1	0%	1,106	4,218	+281%	0	0	—
case-06 During an audit of device Edge-1, BGP neighbor 192.168.12.2 shows a status of 'Idle (Admin)'. What does this specific status string mean, and what testbed command inspects BGP neighbor state?	fail→pass	6,058	4,915	-19%	1	1	0%	1,062	4,016	+278%	0	0	—
case-07 In an EIGRP topology table on device Dist-R1, several routes are marked with letter 'A' while neighbor queue counts are non-zero. What does state 'A' signify, and what operational risk does it present if unconsumed?	pass→pass	10,601	7,152	-33%	1	1	0%	1,813	4,273	+136%	0	0	—
case-08 We need to inspect the forwarding table for the management VRF named 'MGMT' on device Core-1. Which show command execution format should be sent to the device runner?	fail→pass	5,796	3,885	-33%	1	1	0%	1,045	3,760	+260%	0	0	—
case-09 An LSDB dump on device Branch-R1 shows Type 5 External LSAs present inside an area configured as a stub area. What does this anomaly indicate in OSPF troubleshooting?	pass→pass	16,064	15,099	-6%	1	1	0%	2,634	5,750	+118%	0	0	—
case-10 We are troubleshooting static route reaching issues on R1. Which individual sub-commands target static routes and directly connected interface routes separately?	fail→pass	8,769	5,316	-39%	1	1	0%	1,414	3,940	+179%	0	0	—
case-11 When configuring mutual redistribution between OSPF and EIGRP on device Border-GW, what diagnostic step checks whether prefix filtering policies and routing protocol parameters are properly enforcing bounds?	fail→fail	11,268	9,341	-17%	1	1	0%	1,869	4,752	+154%	0	0	—
case-12 Device R1's BGP session to neighbor 10.1.1.2 is trapped in Active state. A junior engineer suggests modifying the local preference. Is this the right troubleshooting direction?	pass→pass	9,715	5,366	-45%	1	1	0%	1,611	4,021	+150%	0	0	—
case-13 How do we check the SPF run counter and OSPF process configuration on router R1 through testbed execution?	fail→pass	13,126	3,653	-72%	1	1	0%	2,598	3,662	+41%	0	0	—
case-14 Prefix 172.16.0.0/16 appears in the BGP table (show ip bgp) on router R2 from an iBGP peer, but it is missing from the main IP routing table (show ip route). What is the primary cause?	pass→pass	8,454	8,807	+4%	1	1	0%	1,650	4,579	+178%	0	0	—
case-15 We need to check whether interface GigabitEthernet0/1 on device R1 is configured as passive or has an unexpected OSPF cost setting. What tool execution should be sent?	fail→pass	5,089	3,623	-29%	1	1	0%	849	3,651	+330%	0	0	—
case-16 After completing a network routing maintenance change on router R1, what command provides a summarized total count of routes per protocol to verify convergence?	fail→pass	4,999	2,940	-41%	1	1	0%	919	3,518	+283%	0	0	—
case-17 In the output of show ip eigrp neighbors on R1, neighbor 10.0.0.2 shows a non-zero Q Count value that stays elevated. What operational issue does this indicate?	pass→pass	9,986	9,124	-9%	1	1	0%	1,696	4,607	+172%	0	0	—
case-20 Please generate the Cisco IOS CLI configuration commands to assign IP address 10.10.10.1/24 to interface GigabitEthernet0/0/1 on router R1 and enable OSPF area 0 on it.	pass→pass	5,317	4,010	-25%	1	1	0%	1,171	3,831	+227%	0	0	—
case-21 Can you write a complete pyATS testbed YAML file defining device R1 with SSH credentials, IP 192.168.1.1, OS type iosxe, and port 22?	pass→pass	5,737	3,775	-34%	1	1	0%	1,186	3,673	+210%	0	0	—
case-22 Switch SW-1 is experiencing a loop on VLAN 10. Please diagnose Spanning Tree Protocol (STP) port states and root bridge placement on SW-1.	pass→fail	12,797	4,824	-62%	1	1	0%	2,274	3,289	+45%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 18 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 18 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 50%
Pass rate with the skill: 77.3%
Cases attempted: 22
Cases counted toward lift: 18
Cases that regressed: 3
Skill version measured: v1
Run completed: 8/16/2026, 12:26 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-04

Prompt

Device R3's OSPF neighbor relationship with R4 is staying in INIT state rather than progressing to FULL. How should we investigate and diagnose this issue on R3 using testbed execution?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run77

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+153% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 19 days ago

Other measured skills in the registry, with their headline benchmark lift.